Madgraph decay

[SOUAD SEMLALI]

Dear MG team,

I am trying to generate events for process gg > h2 > h3 z* > h1 z* z* > bb jj mu+mu-, with h1 > b b~, h2 is. SM-like and mh1, mh3 < mh2 (2HDM type-1).

PS: z's are off-shell, where z* > mu+ mu- and z* > j j. The syntax suggested is : gg> h2 z > h3 mu+ mu-, ( h3 > z > h1 j j, h1 > b b~)

To generate g g > h2, the model is imported (2HDM type-I) , followed by adding the hgg_plugin (after changing the pdg of H from 25 to 35). I have checked the cross section computed by MG, it is \sigma (g g > h2) =17.63pb.

After that, I used MG to compute the cross-section of the process. I did some checks without/with decaying h1 to b b~ (Part 1/part2) ,
Part 1:
  gg> h2 z > h3 mu+ mu-, ( h3 > z > h1 j j)
    cross section = 3.57x10^-4 pb

Part 2 : after decaying h1 > b b~
  gg> h2 z > h3 mu+ mu-, ( h3 > z > h1 j j, h1 > b b~)
        cross section = 0.31pb

with, mh1 = 12.68 GeV, mh2 = 125.00 GeV, mh3 = 68.87 GeV, mhc = 135.00, tb = 6.3, sinbma = -0.16, wh1 = 4.680000e-09, wh2 = 4.520000e-03, wh3 = 1.130000e-04 , l2 = 0.262, l3 = -.592, lr7 = 0.00

The first value (part1) seems more reasonable since h2 is SM-like. After decaying h1 to b b~, the cross section become huge (part2),. It is worth to mention that the total width of h1 is set in the parameter card to 4.68x10^-9 and Br(h1 -> b b~) ~0.69.

Setting the total width of h1 to auto is not working, an error message is appearing. Please find below the error message:

INFO: failed to load model from pickle file. Try importing UFO from File
INFO: load particles
INFO: load vertices
INFO: Change particles name to pass to MG5 convention
INFO: syntax: compute_widths Particle [Particles] [OPTIONS]
INFO: -- Compute the widths for the particles specified.
INFO: By default, this takes the current param_card and overwrites it.
INFO: Precision allows to define when to include three/four/... body decays (LO).
INFO: If this number is an integer then all N-body decay will be included.
INFO: Various options:

INFO: --body_decay=X: Parameter to control the precision of the computation
INFO: if X is an integer, we compute all channels up to X-body decay.
INFO: if X <1, then we stop when the estimated error is lower than X.
INFO: if X >1 BUT not an integer, then we X = N + M, with M <1 and N an integer
INFO: We then either stop at the N-body decay or when the estimated error is lower than M.
INFO: default: 4.0025
INFO: --min_br=X: All channel which are estimated below this value will not be integrated numerically.
INFO: default: precision (decimal part of the body_decay options) divided by four
INFO: --precision_channel=X: requested numerical precision for each channel
INFO: default: 0.01
INFO: --path=X: path for param_card
INFO: default: take value from the model
INFO: --output=X: path where to write the resulting card.
INFO: default: overwrite input file. If no input file, write it in the model directory
INFO: --nlo: Compute NLO width [if the model support it]
ERROR: Invalid command: 25 is not a valid argument for compute_widths
INFO: Update the dependent parameter of the param_card.dat
Command "generate_events run_01" interrupted with error:
ValueError : invalid literal for int() with base 10: 'autoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoautoaut
Please report this bug on https://bugs.launchpad.net/mg5amcnlo
More information is found in 'ME5_debug'.
Please attach this file to your report.

I did some checks:
  - No problem after using the model file without hgg plugin and setting the width(s) to auto for h1 and/or h2 (good values)

   No problem after using the model file with hgg plugin and setting the width to auto for h2 (good values)

    Error message mentioned above, after using the model file with hgg plugin and setting the width to auto for h1.

 Many thanks for any suggestions.

 Kind regards

[SOUAD SEMLALI (souad-sem)]

 Please find attached the model file

[Olivier Mattelaer (olivier-mattelaer)]

You will find the patch to the add model command here.
Note that you have to either removed your merged model first to have this to applied or modified the file particles.py and change particle 9000006 to

h1__1 = Particle(pdg_code = 9000006,
                 name = 'h1__1',
                 antiname = 'h1__1',
                 spin = 1,
                 color = 1,
                 mass = Param.MP,
                 width = Param.WH1_big,
                 texname = 'h1',
                 antitexname = 'h1',
                 charge = 0.0,
                 line = 'dashed',
                 propagating = True,
                 goldstoneboson = False,
                 GhostNumber = 0,
                 LeptonNumber = 0,
                 selfconjugate = True)

Cheers,

Olivier

[Olivier Mattelaer (olivier-mattelaer)]

By the way do remember that loop-induced decay are NOT included in the computation of the total-width this will likely lead to a bias in the total width and to your final cross-section.

Cheers,

Olivier

[SOUAD SEMLALI (souad-sem)]

  Dear Olivier,

  Thanks a lot for your help!. The bug is finally fixed. After setting wh1 to auto, the total width of h1 is computed (~2.5x10^-9 GeV) and Br(h1 > b b~)~73% which is in an agreement with 2hdmc; Although that the loop-induced decays are not included in the total width.

However, using MG:
 \sigma(gg > h2 z > h3 mu+ mu-, ( h3 > z > h1 j j, h1 > b b~)) = 0.59pb; which is big value.

I have checked the case where h1 is decaying instead into \tau+ \tau-:
  \sigma(gg > h2 z > h3 mu+ mu-, ( h3 > z > h1 j j, h1 > \tau+ \tau-)) = 3.54x10^-5 pb; which is acceptable.

How can I overcome the overestimation of the cross-section in the case where h1 is decaying to bb?

On the other hand, using both MG and Madspin gives a different cross section:
   MG: \sigma(gg > h2 z > h3 mu+ mu-) = 6.26x10^-4pb
followed by madspin to decay h3 to :
   decay h3 > z > h1. j j h1 > b b~
the cross section becomes 2.85x10^-5

How can we explain this difference between MG and MG+MS?

Many thanks for any suggestions.

 Thanks again
 Best regards

[Olivier Mattelaer (olivier-mattelaer)]

Are you sure that you set correctly the benchmark (and in particular the width) for your first run?

Olivier

[SOUAD SEMLALI (souad-sem)]

 Dear Olivier,

  Many thanks for your reply. I did set correctly the width of h1, h2 and h3 in the parameter card. Even after setting the width of h1 to auto, the cross-section is still big. Please find below the width from the parameter card after the run.

# PDG Width
DECAY 25 2.507258e-09
# BR NDA ID1 ID2 ...
   7.306541e-01 2 5 -5 # 1.8319382895272787e-09
   1.659561e-01 2 4 -4 # 4.1609462719106996e-10
   1.018637e-01 2 15 -15 # 2.553985133493676e-10
   1.115686e-03 2 3 -3 # 2.7973120534868743e-12
   4.069898e-04 2 13 -13 # 1.0204283485242734e-12
   2.779234e-06 2 1 -1 # 6.9682572737808224e-15
   7.114506e-07 2 2 -2 # 1.7837901895657866e-15
   9.523247e-09 2 11 -11 # 2.387723627205521e-17
#
# PDG Width
DECAY 35 4.520000e-03
#
# PDG Width
DECAY 36 1.130000e-04
#
# PDG Width
DECAY 37 1.000000e+00
#
# PDG Width
DECAY 9000006 5.753088e-03
</slha>
<MGGenerationInfo>
# Number of Events : 10000
# Integrated weight (pb) : 0.5919814999999999

Following the previous email, I have modified the file particles.py

h1__1 = Particle(pdg_code = 9000006,
                 name = 'h1__1',
                 antiname = 'h1__1',
                 spin = 1,
                 color = 1,
                 mass = Param.MP,
                 width = Param.WH1_big,
                 texname = 'h1',
                 antitexname = 'h1',
                 charge = 0.0,
                 line = 'dashed',
                 propagating = True,
                 goldstoneboson = False,
                 GhostNumber = 0,
                 LeptonNumber = 0,
                 selfconjugate = True)

and parameter.py:
WH1_big = Parameter(name = 'WH1_big',
                nature = 'external',
                type = 'real',
                value = 0.00575308848,
                texname = '\\text{WH1}',
                lhablock = 'DECAY',
                lhacode = [9000006])

Many thanks for any suggestions and help.

Best regards

[Olivier Mattelaer (olivier-mattelaer)]

Can you start another bug report for that and include your param_card and run_card such that I can reproduce your number.

Thanks,

Olivier