Poles do not cancel in aMC@NLO 2.2.1

Hi Luca,
so, the poles miscancleation looks like due to an instability related to the small charm mass.
I have informed some MG5_aMC collaborators, they are looking into it.

For what concerns the compiler, i have some questions to you
1) at the end of the output command, it should be printed on screen something like 'use c++ compiler COMPILER'
2) what happens if you type in the shell COMPILER --version (with COMPILER what you get in the previous step)?
3) Can you check inside Source/make_opts, in the directory which has been exported, what is the value for CFLAGS and CXXFLAGS?

Thanks a lot,

Marco

Hi Marco,

thanks for your reply.

1) I use Apple LLVM version 6.0 (clang-600.0.51) (based on LLVM 3.5svn).

2)The environmental variables are defined as

ifeq ($(origin CXXFLAGS),undefined)
  #CXXFLAGS= -g
  CXXFLAGS = -O
endif

ifeq ($(origin CFLAGS),undefined)
  #CFLAGS= -g
  CFLAGS = -O
endif

However, as I wrote in the bug section, the problem is reproducible on different machines which have different compilers, in particular on a CERN machine with SLC6.5 with gcc4.8, on a linux machine with ubuntu (and I think gcc4.8, but I can check), and on another mac.

Thanks

Luca

Hi Marco,

I think I misunderstood your questions about the compiler. The problem related to libstc++ affects only my mac, everything is fine on the other machines.

Thanks

Luca

Hi Luca,
so the message ‘Using c++ compiler …’ points to your clang?

On 29 Oct 2014, at 12:46, Luca <email address hidden> wrote:

> Question #256309 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/256309
>
> Luca posted a new comment:
> Hi Marco,
>
> I think I misunderstood your questions about the compiler. The problem
> related to libstc++ affects only my mac, everything is fine on the other
> machines.
>
> Thanks
>
> Luca
>
> --
> You received this question notification because you are a direct
> subscriber of the question.

Hi Luca,
so for the poles, just change
      energy = max((ebeam(1)+ebeam(2))/2.0d0,2.0d0*totmass)

with
      energy = max((ebeam(1)+ebeam(2))/20.0d0,2.0d0*totmass)

inside check_poles.f

For the compilation issue, can you try to type

set cpp_compiler clang

in the madgraph interface and then re-export the process (the CFLAGS should now be different inside make_opts)

Let me know,

Marco

On 29 Oct 2014, at 13:31, Marco Zaro <email address hidden> wrote:

> Hi Luca,
> so the message ‘Using c++ compiler …’ points to your clang?
>
> On 29 Oct 2014, at 12:46, Luca <email address hidden> wrote:
>
>> Question #256309 on MadGraph5_aMC@NLO changed:
>> https://answers.launchpad.net/mg5amcnlo/+question/256309
>>
>> Luca posted a new comment:
>> Hi Marco,
>>
>> I think I misunderstood your questions about the compiler. The problem
>> related to libstc++ affects only my mac, everything is fine on the other
>> machines.
>>
>> Thanks
>>
>> Luca
>>
>> --
>> You received this question notification because you are a direct
>> subscriber of the question.
>

Hi Marco

The poles cancels correctly with the change you proposed. However another error appears:

Traceback (most recent call last):
  File "/Users/rottoli/Documents/Physics/computing/MG5_aMC_v2_2_1_bug/madgraph/interface/extended_cmd.py", line 879, in onecmd
    return self.onecmd_orig(line, **opt)
  File "/Users/rottoli/Documents/Physics/computing/MG5_aMC_v2_2_1_bug/madgraph/interface/extended_cmd.py", line 872, in onecmd_orig
    return func(arg, **opt)
  File "/Users/rottoli/Documents/Physics/computing/MG5_aMC_v2_2_1_bug/madgraph/interface/amcatnlo_run_interface.py", line 1190, in do_launch
    evt_file = self.run(mode, options)
  File "/Users/rottoli/Documents/Physics/computing/MG5_aMC_v2_2_1_bug/madgraph/interface/amcatnlo_run_interface.py", line 1565, in run
    self.cross_sect_dict = self.read_results(output, mode)
  File "/Users/rottoli/Documents/Physics/computing/MG5_aMC_v2_2_1_bug/madgraph/interface/amcatnlo_run_interface.py", line 1727, in read_results
    'Please check the .log files inside the directories which failed.')
aMCatNLOError: An error occurred during the collection of results.

Furthermore, the compilation issues has not been resolved. I think that the problem might be related to fastjet which is not compiled with lstc++ (see https://github.com/davidchall/homebrew-hep/issues/26) on a mac (I installed fast jet both via homebrew and by myself in order to check).

Luca

Ciao Luca,
before, you should have a list of the processes which fail, or somethign like that…
Can you send me the list and check the log.txt inside these folders?
Marco

On 29 Oct 2014, at 14:36, Luca <email address hidden> wrote:

> Question #256309 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/256309
>
> Status: Answered => Open
>
> Luca is still having a problem:
> Hi Marco
>
> The poles cancels correctly with the change you proposed. However
> another error appears:
>
> Traceback (most recent call last):
> File "/Users/rottoli/Documents/Physics/computing/MG5_aMC_v2_2_1_bug/madgraph/interface/extended_cmd.py", line 879, in onecmd
> return self.onecmd_orig(line, **opt)
> File "/Users/rottoli/Documents/Physics/computing/MG5_aMC_v2_2_1_bug/madgraph/interface/extended_cmd.py", line 872, in onecmd_orig
> return func(arg, **opt)
> File "/Users/rottoli/Documents/Physics/computing/MG5_aMC_v2_2_1_bug/madgraph/interface/amcatnlo_run_interface.py", line 1190, in do_launch
> evt_file = self.run(mode, options)
> File "/Users/rottoli/Documents/Physics/computing/MG5_aMC_v2_2_1_bug/madgraph/interface/amcatnlo_run_interface.py", line 1565, in run
> self.cross_sect_dict = self.read_results(output, mode)
> File "/Users/rottoli/Documents/Physics/computing/MG5_aMC_v2_2_1_bug/madgraph/interface/amcatnlo_run_interface.py", line 1727, in read_results
> 'Please check the .log files inside the directories which failed.')
> aMCatNLOError: An error occurred during the collection of results.
>
> Furthermore, the compilation issues has not been resolved. I think that
> the problem might be related to fastjet which is not compiled with
> lstc++ (see https://github.com/davidchall/homebrew-hep/issues/26) on a
> mac (I installed fast jet both via homebrew and by myself in order to
> check).
>
> Luca
>
> --
> You received this question notification because you are a direct
> subscriber of the question.

Hi Marco,
The problem with the generation of the events happens also on a linux machine. This message pops up

1 jobs did not terminate correctly:
P0_gg_ccx/GF1
The results are probably not correct. Please check the relevant log files corresponding to the above jobs for more information.
Found 4 correctly terminated jobs

the last lines in the log file are

 alpha_s value used for the virtuals is (for the first PS point): 0.17741006015248523
 ---- POLES CANCELLED ----
 Cannot do MC over hel: "HelFilter.dat" does not exist or does not have the correct format. Change NHelForMCoverHels in FKS_params.dat to explicitly summ over them instead.
Time in seconds: 0

I had a look to FKS_params.dat, but I don't know how to change it...

Ciao Luca,
I have asked my colalborators about this...
it should not be related to FKSparams, as soon as i get an answer i'll let you know
Cheers,

Marco

Ciao Marco,

Thanks for your answer. Do you suggest to work on the old version 2.1.2 of amc@nlo or to wait for a bug fix? In particular, do you think that the results obtained with 2.1.2 can be different from the results one would get with 2.2, with respect to this particular process? (btw, I would like to shower the results with Pythia 8.2, so maybe it could be better to wait for a future update which supports the shower inside amc@nlo which at the moment is possibile only with pythia 8.1)

Many thanks

Luca

Ciao Luca,
in the meanwhile, please work with 2.1.2, the results should be the same.
We are currently working on the inclusion of the possibility of py8.2, it should be available in the next release.
Thanks,

Marco

On 30 Oct 2014, at 17:51, Luca <email address hidden> wrote:

> Question #256309 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/256309
>
> Luca posted a new comment:
> Ciao Marco,
>
> Thanks for your answer. Do you suggest to work on the old version 2.1.2
> of amc@nlo or to wait for a bug fix? In particular, do you think that
> the results obtained with 2.1.2 can be different from the results one
> would get with 2.2, with respect to this particular process? (btw, I
> would like to shower the results with Pythia 8.2, so maybe it could be
> better to wait for a future update which supports the shower inside
> amc@nlo which at the moment is possibile only with pythia 8.1)
>
> Many thanks
>
> Luca
>
> --
> You received this question notification because you are a direct
> subscriber of the question.

Ciao Marco,

thanks for your advice. Let me know if there are news about my problem with charm production in 2.2.1
thanks

Luca

Thanks marco zaro, that solved my question.