An error occurred during the collection of results

Asked by hek

Dear MadGraph team,

I'm using aMC@NLO to generate events for the processes:

g gamma -> ttg [real=QCD] @ SM

There is something wrong at the end:

Error detected in "launch "
write debug file /home/hek/work/MG5_aMC_v2_5_0_beta2/gy/run_02_tag_1_debug.log
If you need help with this issue please contact us on https://answers.launchpad.net/mg5amcnlo
aMCatNLOError : An error occurred during the collection of results.
 Please check the .log files inside the directories which failed:
 /home/hek/work/MG5_aMC_v2_5_0_beta2/gy/SubProcesses/P0_ga_ttxg/GF5/log.txt
 /home/hek/work/MG5_aMC_v2_5_0_beta2/gy/SubProcesses/P0_ga_ttxg/GF8/log.txt

The debug log is:

launch
Traceback (most recent call last):
  File "/home/hek/work/MG5_aMC_v2_5_0_beta2/gy/bin/internal/extended_cmd.py", line 1393, in onecmd
    return self.onecmd_orig(line, **opt)
  File "/home/hek/work/MG5_aMC_v2_5_0_beta2/gy/bin/internal/extended_cmd.py", line 1348, in onecmd_orig
    return func(arg, **opt)
  File "/home/hek/work/MG5_aMC_v2_5_0_beta2/gy/bin/internal/amcatnlo_run_interface.py", line 1209, in do_launch
    evt_file = self.run(mode, options)
  File "/home/hek/work/MG5_aMC_v2_5_0_beta2/gy/bin/internal/amcatnlo_run_interface.py", line 1438, in run
    jobs_to_collect,mint_step,mode,mode_dict[mode],fixed_order=False)
  File "/home/hek/work/MG5_aMC_v2_5_0_beta2/gy/bin/internal/amcatnlo_run_interface.py", line 1645, in collect_the_results
    self.append_the_results(jobs_to_run,integration_step)
  File "/home/hek/work/MG5_aMC_v2_5_0_beta2/gy/bin/internalValue of current Options:
Value of current Options:
              text_editor : None
              web_browser : None
        cluster_temp_path : None
                  timeout : 60
       cluster_local_path : /cvmfs/cp3.uclouvain.be/madgraph/
            cluster_queue : None
         madanalysis_path : None
                   lhapdf : /home/hek/program/LHAPDF/bin/lhapdf-config
             cluster_size : 100
           cluster_memory : None
                    pjfry : None
    cluster_status_update : (600, 30)
             cluster_time : None
            f2py_compiler : None
               hepmc_path : None
             pythia8_path : None
                hwpp_path : None
   automatic_html_opening : False
       cluster_retry_wait : 300
             stdout_level : None
          pythia-pgs_path : None
                 mg5_path : /home/hek/work/MG5_aMC_v2_5_0_beta2
                  td_path : None
             delphes_path : None
              thepeg_path : None
             cluster_type : condor
        madanalysis5_path : None
      exrootanalysis_path : None
         fortran_compiler : None
                  nb_core : 4
              auto_update : 7
         cluster_nb_retry : 1
               eps_viewer : None
             syscalc_path : None
                  fastjet : None
                    golem : None
             cpp_compiler : None
      notification_center : True
                 run_mode : 2

I try to change the value of 'tiny' in the file ' .Template/NLO/SubProcesses/montecarlocounter.f ' from 1d-4 to 1d-2 & 1d-6 to 1d-4. But it doesn't work.

Then I go to the ./SubProcesses/P0_ga_ttxg/GF5 directory and run ../madevent_mintMC <input_app.txt as https://answers.launchpad.net/mg5amcnlo/+question/293409 said. It gives error like this:

 Warning in get_cms_costh_fks 1.0591371900941133
 Warning in get_cms_costh_fks 1.1615514652537993
 Warning in get_cms_costh_fks 1.0058992492042804
 Warning in get_cms_costh_fks 1.0128619148172786
 Error 3 in kinematics_driver: negtive sqrt
                 -Infinity 99
 Error 3 in kinematics_driver: negtive sqrt
                 -Infinity 100

It seems there are something wrong in phase-space decomposition. What should I do next?

Many thanks,
hek

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
marco zaro Edit question
Solved by:
marco zaro
Solved:
Last query:
Last reply:
Revision history for this message
Valentin Hirschi (valentin-hirschi) said :
#1

At NLO, MG5aMC only supports processes with *both* color singlets or *both* non-color singlets in the initial state (so p p or e+ e- basically).

Therefore, Your process cannot be computed in MG5aMC as of now. Also, notice that including only the born and real corrections (like you did with real=QCD) is unphysical, and when doing this you must provide the virtual contribution by yourself.
You normally alway want to use [QCD] (or equivalently, [all=QCD]).

Revision history for this message
hek (mitsuha) said :
#2

Dear Hirschi:

         Thanks for your reply.

          I am using [real=QCD] just for a test. The virtual part can be generated and ran sucessfully by aMc@NLO.

          You say that aMc@NLO cannot handle ep or gamma p processes right now. But I also use aMc@NLO to compute some processes like: g gamma -> tt & u gamma -> ttu. The software can run and generate events without any error. So can I use these events for further work?

         Thanks

          hek

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#3

Hi,

> g gamma -> tt & u gamma -> ttu.

We are indeed able to do such type of computation at LO (that’s what you mean I guess).
And those events are valid. As usual you have to be careful with the photon pdf and check carefully which type of
contribution are include since it the PDF dependent.

Cheers,

Olivier

> On Oct 14, 2016, at 19:13, hek <email address hidden> wrote:
>
> Question #403009 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/403009
>
> hek posted a new comment:
> Dear Hirschi:
>
> Thanks for your reply.
>
> I am using [real=QCD] just for a test. The virtual part can be
> generated and ran sucessfully by aMc@NLO.
>
> You say that aMc@NLO cannot handle ep or gamma p processes
> right now. But I also use aMc@NLO to compute some processes like: g
> gamma -> tt & u gamma -> ttu. The software can run and generate events
> without any error. So can I use these events for further work?
>
> Thanks
>
> hek
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
hek (mitsuha) said :
#4

Dear Mattelaer:

> g gamma -> tt & u gamma -> ttu.

        The processes above are ran and generated events at NLO.

        Thanks for your reply; I'm sorry for I forget this information.

         Many thanks

         hek

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#5

What is the command that you use?
I guess that you are mistaking something since we are not able to handle gluon photon initial state at NLO

Cheers,

Olivier
> On Oct 15, 2016, at 07:57, hek <email address hidden> wrote:
>
> Question #403009 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/403009
>
> hek posted a new comment:
> Dear Mattelaer:
>
>> g gamma -> tt & u gamma -> ttu.
>
> The processes above are ran and generated events at NLO.
>
> Thanks for your reply; I'm sorry for I forget this information.
>
> Many thanks
>
> hek
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
hek (mitsuha) said :
#6

Dear Mattelaaer:

            I just test ga -> tt process in my computer and it works ok.

           The command I use shows below:

            1. generant g a > t t~ [all=QCD]
            2. output gatt
            3. launch gatt
            4. set shower=OFF
            5. edit the param_card.dat & run_card.dat to set energy to 7000; set PDF to lhapdf 244600; set scale to 173GeV( both MuR and MuF, they are both fixed); Disable reweight_scale. Other parameters are default.

           The result is:

   --------------------------------------------------------------
      Summary:
      Process g a > t t~ [all=QCD]
      Run at p-p collider (7000.0 + 7000.0 GeV)
      Number of events generated: 10000
      Total cross section: 3.578e+00 +- 2.4e-02 pb
   --------------------------------------------------------------

          Best,
          hek

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#7

Hi Hek,

For NLO processes, you should never mention the savour of the jet/proton.
(you have a serious warning when you do not do it)
So the correct syntax should be
generate p a > t t~ [QCD]
I have checked that in this particular case, you have the exact same diagram so it should be fine.
I will assign this thread to Marco just to be 100% since Valentin claims that this should not be ok.

On the other hand, I would be quite suspicious about event generated with
 u gamma -> ttu
you should really pass to
generate p a > t t~ j [QCD]

Cheers,

Olivier

Revision history for this message
hek (mitsuha) said :
#8

Dear Olivier:

       Thanks for your check.

       The process p a > ttj contains g a > ttg subprocess; so I think the aMc@NLO will obtain same errors at end.

        I will generate p a > ttj[real] for a test since this process runs not quickly and the virtual part seems to have no problem.

        Best,

        hek

Revision history for this message
hek (mitsuha) said :
#9

Dear Olivier::

       Somethings look interesting. The aMc@NLO can run and generate events of process p a > ttj[real] in a couple of minutes:

   --------------------------------------------------------------
      Summary:
      Process p a > t t~ j[real=QCD]
      Run at p-p collider (7000.0 + 7000.0 GeV)
      Number of events generated: 10000
      Total cross section: 8.118e-01 +- 1.5e-02 pb
   --------------------------------------------------------------

      And I have checked the subprcoess that g a > ttg[real] is contained.

      I will generate p a > ttj[all=QCD] next. If the aMc@NLO can run and get result without any error, how can I confirm this result is right?

    Best,

    hek

Revision history for this message
Best marco zaro (marco-zaro) said :
#10

Hi Hek,
this kind of process need some care, as it was mentioned by Valentin. The problem is, QCD corrections to a g > t t~ j are of the same perturbative order as EW corrections to p p > t t~ j. In particular, with the current version of the code, you will compute all real emissions which can be obtained starting from a g > t t~ j and splitting all particles with only QCD splitting. you will not get, for example, u g > t t~ u g where one includes the a > u u~ splitting in the initial state.
So, despite the fact that you get a result, it may not be the correct one...
What are you exactly interested in?
Best,

Marco

Revision history for this message
hek (mitsuha) said :
#11

Dear Marco:

       Yes, I'm working on EW corrections to p p > t t~j process. I have carefully checked and make sure that all subprocesses are included in my work( such as a > u u~ splitting processes you mentioned above). The g a > t t~ g [QCD] process has been calculated before but the result is too worse. So I want to use MadGraph to generate and obtain reuslt of this single subprcoess and other tools to deal with remains.

       Thanks for everyone help me. I think the problem has been solved.

        Best,

        hek

Revision history for this message
hek (mitsuha) said :
#12

Thanks marco zaro, that solved my question.