Run MadSpin without changing event weights?

Asked by Mike Hance on 2019-09-23

Dear experts,

I'd like to use MadSpin to decay some SUSY particles in a simplified model where the branching ratio is artificially set to 100% for certain channels. I produce the un-decayed events with MadGraph, and get a reasonable cross section. When MadSpin decays the heavy states, the event weights become very small, since the decay I'm interested in happens to be suppressed given the param_card used for event generation. I then need to unfold this scale factor to get back to my simplified model after showering. This isn't the end of the world, but it's mildly annoying.

Is it possible to ask MadSpin (via a flag?) to not modify the event weights provided in the LHE file, and just provide the decay kinematics? In this case, that's all I need it for.

Thanks!

-Mike

Question information

Language:
English Edit question
Status:
Answered
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Last query:
2019-09-23
Last reply:
2019-09-24

Did you provide your BR in the param_card?
If you provide it to 100% (and provide the BR for all particle written in the madspin_card)
then the weight will be multipllied by one (i.e. the BR)

Cheers,

Olivier

> On 23 Sep 2019, at 20:43, Mike Hance <email address hidden> wrote:
>
> New question #684169 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/684169
>
> Dear experts,
>
> I'd like to use MadSpin to decay some SUSY particles in a simplified model where the branching ratio is artificially set to 100% for certain channels. I produce the un-decayed events with MadGraph, and get a reasonable cross section. When MadSpin decays the heavy states, the event weights become very small, since the decay I'm interested in happens to be suppressed given the param_card used for event generation. I then need to unfold this scale factor to get back to my simplified model after showering. This isn't the end of the world, but it's mildly annoying.
>
> Is it possible to ask MadSpin (via a flag?) to not modify the event weights provided in the LHE file, and just provide the decay kinematics? In this case, that's all I need it for.
>
> Thanks!
>
> -Mike
>
>
>
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Mike Hance (mhance) said : #2

Hi Olivier,

Yes, I've provided the weight in the param card (for other reasons), but I was under the impression that MadSpin is recomputing the BR from the model? I've read the following thread in the past, and assumed that MadGraph and MadSpin were doing the same thing here:

https://answers.launchpad.net/mg5amcnlo/+question/253601

Thanks!

-Mike

if you have the BR provided in the param_card for all particle with the line
decay X
in the madspin_card, then MadSpin does not recompute the BR.

Cheers,

Olivier

> On 23 Sep 2019, at 20:57, Mike Hance <email address hidden> wrote:
>
> Question #684169 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/684169
>
> Mike Hance posted a new comment:
> Hi Olivier,
>
> Yes, I've provided the weight in the param card (for other reasons), but
> I was under the impression that MadSpin is recomputing the BR from the
> model? I've read the following thread in the past, and assumed that
> MadGraph and MadSpin were doing the same thing here:
>
> https://answers.launchpad.net/mg5amcnlo/+question/253601
>
> Thanks!
>
> -Mike
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Mike Hance (mhance) said : #4

Hi Olivier,

Thanks for bearing with me while I try to understand this better.

I'm producing chargino pairs. Here's the block from the param card in the LHE file... filled with dummy BR's while I try to understand what's going on:

# PDG Width
DECAY 1000024 1.70414503E-02 # chargino1+ decays
# BR NDA ID1 ID2 ID3
     0.00000000E+00 2 1000022 24 # BR(~chi_1+ -> ~chi_10 W+)
     0.10000000E+00 3 1000022 -11 12 # BR(~chi_1+ -> ~chi_10 e+ nu_e)
     0.10000000E+00 3 1000022 -13 14 # BR(~chi_1+ -> ~chi_10 mu+ nu_mu)
     0.10000000E+00 3 1000022 -15 16 # BR(~chi_1+ -> ~chi_10 tau+ nu_tau)
     0.35000000E+00 3 1000022 2 -1 # BR(~chi_1+ -> ~chi_10 u db)
     0.35000000E+00 3 1000022 4 -3 # BR(~chi_1+ -> ~chi_10 c sb)

Here's the beginning of the LHE (minus headers) from before MadSpin runs:

<init>
2212 2212 6.500000e+03 6.500000e+03 0 0 247000 247000 -3 1
   +5.0921433e-02 +2.4494551e-04 +1.2730361e-05 1
<generator name='MadGraph5_aMC@NLO' version='2.6.6'>please cite 1405.0301 </generator>
</init>
<event>
 6 1 +1.2730358e-05 2.54190100e+02 7.81653100e-03 1.11068500e-01
        2 -1 0 0 502 0 +0.0000000000e+00 +0.0000000000e+00 +7.9682434100e+02 7.9682434100e+02 0.0000000000e+00 0.0000e+00 -1.0000e+00
        2 -1 0 0 501 0 -0.0000000000e+00 -0.0000000000e+00 -4.9353227357e+02 4.9353227357e+02 0.0000000000e+00 0.0000e+00 -1.0000e+00
  1000024 1 1 2 0 0 +8.5974788432e+01 -1.2656988176e+01 +9.5755989026e+00 1.3282904686e+02 1.0000000000e+02 0.0000e+00 1.0000e+00
 -1000024 1 1 2 0 0 -2.7527004466e+01 +7.1273143542e+00 +2.2124894348e+01 1.0629226470e+02 1.0000000000e+02 0.0000e+00 1.0000e+00
        2 1 1 2 501 0 -1.0864617453e+02 +2.4244936151e+01 +6.5485324857e+02 6.6424738263e+02 0.0000000000e+00 0.0000e+00 -1.0000e+00
        2 1 1 2 502 0 +5.0198390567e+01 -1.8715262329e+01 -3.8326167439e+02 3.8698792038e+02 0.0000000000e+00 0.0000e+00 -1.0000e+00
</event>

The "+5.0921433e-02" in the init block corresponds to the 50 fb cross section that MadGraph (2.6.6) reports in log files. The "+1.2730358e-05" in the event block is just the cross section divided by 4000 (the number of events in the run). So far so good.

After running MadSpin:

<init>
2212 2212 6.500000e+03 6.500000e+03 0 0 247000 247000 -3 1
   +1.7013555e-25 +8.1839684e-28 +4.2533897e-29 1
<generator name='MadGraph5_aMC@NLO' version='2.6.6'>please cite 1405.0301 </generator>
</init>
<event>
12 1 +4.2533887e-29 2.54190100e+02 7.81653100e-03 1.11068500e-01
        2 -1 0 0 502 0 +0.0000000000e+00 +0.0000000000e+00 +7.9682434100e+02 7.9682434100e+02 0.0000000000e+00 0.0000e+00 -1.0000e+00
        2 -1 0 0 501 0 -0.0000000000e+00 -0.0000000000e+00 -4.9353227357e+02 4.9353227357e+02 0.0000000000e+00 0.0000e+00 -1.0000e+00
  1000024 2 1 2 0 0 +8.5974788432e+01 -1.2656988176e+01 +9.5755989026e+00 1.3282904686e+02 1.0000000000e+02 0.0000e+00 0.0000e+00
 -1000024 2 1 2 0 0 -2.7527004466e+01 +7.1273143542e+00 +2.2124894348e+01 1.0629226470e+02 1.0000000000e+02 0.0000e+00 0.0000e+00
        2 1 1 2 501 0 -1.0864617453e+02 +2.4244936151e+01 +6.5485324857e+02 6.6424738263e+02 0.0000000000e+00 0.0000e+00 -1.0000e+00
        2 1 1 2 502 0 +5.0198390567e+01 -1.8715262329e+01 -3.8326167439e+02 3.8698792038e+02 0.0000000000e+00 0.0000e+00 -1.0000e+00
        2 1 3 3 503 0 +1.0423758674e+00 +8.6819078501e-02 -2.6653141661e-01 1.0794090963e+00 0.0000000000e+00 0.0000e+00 -1.0000e+00
       -1 1 3 3 0 503 +4.0145187771e-01 -4.2361974336e-01 -1.6619615564e-01 6.0682654789e-01 0.0000000000e+00 0.0000e+00 1.0000e+00
  1000022 1 3 3 0 0 +8.4530960688e+01 -1.2320187511e+01 +1.0008326475e+01 1.3114281122e+02 9.9000000000e+01 0.0000e+00 1.0000e+00
        1 1 4 4 504 0 -4.6135030897e-01 -9.3007669779e-02 +2.0706199145e-02 4.7108733894e-01 0.0000000000e+00 0.0000e+00 -1.0000e+00
       -2 1 4 4 0 504 -4.9196852323e-02 -1.1919761438e-01 -4.7346161948e-01 4.9070796475e-01 0.0000000000e+00 0.0000e+00 1.0000e+00
  1000022 1 4 4 0 0 -2.7016457304e+01 +7.3395196383e+00 +2.2577649768e+01 1.0533046939e+02 9.9000000000e+01 0.0000e+00 -1.0000e+00
</event>

Now the cross section is "+1.7013555e-25", and the event weights are modified accordingly.

This is running MadGraph 2.6.6, but we've been seeing this same behavior for ages, so I assumed it was expected/correct. Am I getting something wrong?

Thanks!

-Mike

Do I understand correctly that you are doing three body decay?
Is this correct?

In that case this is not the "normal" madspin but one of the special mode (the default mode of madspin does not support three body decay).

So the first question is which special mode are you using?
In the spinmode=none then indeed since this is a stupid "glue mode" we start by generating events for the 1 to 3 body decay and compute the partial width associated (which allow to have some simple cuts/...) and therefore the solution would be to put the width computed during that stage as the total width of that particle.

Cheers,

Olivier

On 24 Sep 2019, at 06:13, Mike Hance <<email address hidden><mailto:<email address hidden>>> wrote:

 1000022 1 3 3 0 0 +8.4530960688e+01 -1.2320187511e+01 +1.0008326475e+01 1.3114281122e+02 9.9000000000e+01 0.0000e+00 1.0000e+00

Mike Hance (mhance) said : #6

Hi Olivier,

Yes, exactly, this is spinmode=none. Sorry for not specifying this earlier!

Where can I find the width computed during the 1->3 stage? I don't see it in the screen output, and it doesn't seem to end up in any of the param cards.

But either way, could it be optional to discard the recomputed partial width in favor of the one from the param card?

Thanks!

-Mike

> Where can I find the width computed during the 1->3 stage? I don't see
> it in the screen output, and it doesn't seem to end up in any of the
> param cards.

No that's correct, you can do the computation manually to have that value.

> But either way, could it be optional to discard the recomputed partial
> width in favor of the one from the param card?

I will not add such options (they are already too many special cases and hidden option within madspin to add additional special cases. Especially in this case, this is something that is easy to fix/handle as post-processing (and actually it is common for analysis to take the cross-section as external input and discard the one from the sample).

Now you can try to use the hidden option
set cross_section xxxxx

This options was designed for hepmc decay file, but you can try to use in this case and if it works it might be equivalent to what you are asking for.

Cheers,

Olivier

> On 24 Sep 2019, at 14:58, Mike Hance <email address hidden> wrote:
>
> Question #684169 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/684169
>
> Mike Hance posted a new comment:
> Hi Olivier,
>
> Yes, exactly, this is spinmode=none. Sorry for not specifying this
> earlier!
>
> Where can I find the width computed during the 1->3 stage? I don't see
> it in the screen output, and it doesn't seem to end up in any of the
> param cards.
>
> But either way, could it be optional to discard the recomputed partial
> width in favor of the one from the param card?
>
> Thanks!
>
> -Mike
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Mike Hance (mhance) said : #8

Hi Olivier,

OK, thanks. I'll try "set cross_section", since this does seem to be what I'm interested in.

Cheers,

-Mike

Can you help with this problem?

Provide an answer of your own, or ask Mike Hance for more information if necessary.

To post a message you must log in.