Survey Scanning only

You can set the number of requested events to be 1 which effectively limits
the computation to the survey only.
Keep in mind that this will of course lead to large Monte-Carlo
uncertainties, so make sure that this is what you need.
You can also chose to launch the madevent run interface with:

./bin/mg5
MG5_aMC> launch PROC_OUTPUT_NAME -i

which sends you to another type of interactive interface with a different
prompt, where new commands specific to that process output are available,
such as:

PROC_OUTPUT_NAME> survey

But we haven't provided options to this command to modify the target
accuracy, so if you want better MC accuracy, the simplest is to ask for N
events, knowing that, as a rule of thumbs, the expected relative accuracy
of the x-section will be of order 1/Sqrt(N).

On Fri, Dec 16, 2016 at 10:27 AM, Siba Prasad Das <
<email address hidden>> wrote:

> New question #406820 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/406820
>
> Hi,
>
> Following earlier correspondence about generating events for
> several SLHA files, like this below is working fine.
> ====================================================
> launch ep_eh1j
> ~/slha1.txt
> launch
> ~/slha2.txt
> launch ep_eh1j -i
> print_results --path=./cross.txt --format=short
> =====================================================
>
> Suppose, I would like to have the cross-sections only.
> I tried few options, e.g., replacing "launch" by "survey", but
> not able to solved. It would be nice to know your suggestions
> at your convenience. Thank you.
>
> Best regards,
> Siba Prasad Das
>
>
>
>
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.
>

--
Valentin

Hi Valentin,

Thanks for your prompt message.

If I understand correctly, this also means getting the cross-section with Nevent=1.
This I dont want. Ofcourse as you said 1/sqrt(N), so uncertainty is very high.

May be I should clarify a bit more, the above syntaxes of few lines
is written in a batch file (suggestion from Olivier), and executing the
batch file, "../bin/mg5_aMC batch_file", just below exact process
directory: ep_eh1j. It does the things for launching and then getting
cross-section, generate events according to run_card.dat and
slha1.txt, slha2.txt, etc... as a param_card.

Likewise, if I somehow changed the "launch" by "survey" (in the batch file) with
large number of events, say 100K(in run-card.dat) , and w/o generating the
events, and for all the SLHA files, can we able to find the stable cross-sections.

I have in mind that not all the SLHA_n.txt as large cross-sections for our interest.
First we survey using "survey" (the command name is just like so) and then only
for the interesting one, we will generate events.

With events generation things are allright, but to save CPU time,
first we would like to do "survey"!

If you need more clarification please let me know at your convenience.

Thank you so much.

Best regards,
Siba

Dear Siba,

The precision obtain after the survey is typically of 10% and completely independent of the number of event that you request.
(this entry is not used by the survey command)
One alternative to the solution of Valentin is actually to do the following:

launch ep_eh1j -i
! cp ~/slha1.txt ep_eh1j/Cards/param_card.dat
survey
! cp ~/slha2.txt ep_eh1j/Cards/param_card.dat
survey
print_results --path=./cross.txt --format=short

If you want to have higher accuracy than 10% you can use the "refine" command after calling the survey.

Cheers,

Olivier

Dear Olivier,

Thanks for your messages. Yes, it is working!

Still one thing to solve: The cross-sections for all benchmarks
is not appending, rather it shows for the last param_card.
I tried using, some path/cat/>/ but none is working.

This is not a big problem though, all the informations are there in
the output log in the terminal. So, I save the screen output and
search the "Cross-section" for all benchamrks.

The refine is also working, with specification like "refine 0.01" (for 1%)
Here unlike "launch", we need the !cp command to pipe the param_card.dat.
For launch, however the appending is working with the run_tag (corresponds
to the benchmarks).

It would be nice, if you please let me know how to append the cross-section
(according to the run_tag) at your convenience!

Otherwise it is working using the screen output!

Thank you so much for your help!

Best regards,
Siba

Hi,

I guess that the command that you are looking for is:
print_results —path=XXXXXX —mode=a —format=short
that you type after each call to survey/refine.

Cheers,

Olivier

On Dec 17, 2016, at 17:22, Siba Prasad Das <email address hidden> wrote:

Question #406820 on MadGraph5_aMC@NLO changed:
https://answers.launchpad.net/mg5amcnlo/+question/406820

  Status: Answered => Open

Siba Prasad Das is still having a problem:
Dear Olivier,

Thanks for your messages. Yes, it is working!

Still one thing to solve: The cross-sections for all benchmarks
is not appending, rather it shows for the last param_card.
I tried using, some path/cat/>/ but none is working.

This is not a big problem though, all the informations are there in
the output log in the terminal. So, I save the screen output and
search the "Cross-section" for all benchamrks.

The refine is also working, with specification like "refine 0.01" (for 1%)
Here unlike "launch", we need the !cp command to pipe the param_card.dat.
For launch, however the appending is working with the run_tag (corresponds
to the benchmarks).

It would be nice, if you please let me know how to append the cross-section
(according to the run_tag) at your convenience!

Otherwise it is working using the screen output!

Thank you so much for your help!

Best regards,
Siba

--
You received this question notification because you are an answer
contact for MadGraph5_aMC@NLO.

Dear Olivier,

Thanks. This solved.

Best regards,
Siba