RuntimeError : Info file not found for PDF set 'NNPDF23_lo_as_0130_qed'

Hi,

I guess that you are using lhapdf6.1.6 and that your PDF set is not NATIVELY supported by that version of lhapdf. This is the reason why the python link of lhapdf fails to link to that pdf set.

You might check with lhapdf author, but I guess that it exists a lhapdf command like
lhapdf-config update
exists to update the database and to handle that pdf.

Otherwise, you will need to install one version of lhapdf6 (likely 6.2.1) which support NATIVELY your pdf set.

Cheers,

Olivier

> On 23 Oct 2018, at 17:07, SAUMYEN KUNDU <email address hidden> wrote:
>
> New question #675492 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/675492
>
> Hi,
>
> While generating a certain process, I was getting the following error in Systematics computation :
> -----------------------------------------------------------------------------------------------------------
> INFO: Running Systematics computation
> INFO: Idle: 0, Running: 4, Completed: 0 [ current time: 20h14 ]
> WARNING: program /usr/bin/python -O /home/saumyen/HEP_Package/MG5_aMC_v2_6_1/PROC_sm_12/bin/internal/systematics.py unweighted_events.lhe.gz ./tmp_1_unweighted_events.lhe.gz --mur=0.5,1,2 --muf=0.5,1,2 --pdf=errorset --start_event=2500 --stop_event=5000 --result=./log_sys_1.txt --lhapdf_config=/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/HEPTools/lhapdf6/bin/lhapdf-config launch ends with non zero status: 1. Stop all computation
> WARNING: program /usr/bin/python -O /home/saumyen/HEP_Package/MG5_aMC_v2_6_1/PROC_sm_12/bin/internal/systematics.py unweighted_events.lhe.gz ./tmp_3_unweighted_events.lhe.gz --mur=0.5,1,2 --muf=0.5,1,2 --pdf=errorset --start_event=7500 --stop_event=10000 --result=./log_sys_3.txt --lhapdf_config=/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/HEPTools/lhapdf6/bin/lhapdf-config launch ends with non zero status: 1. Stop all computation
> INFO: Running Systematics computation
> Command "generate_events run_01" interrupted with error:
> RuntimeError : Info file not found for PDF set 'NNPDF23_lo_as_0130_qed'
> Please report this bug on https://bugs.launchpad.net/mg5amcnlo
> More information is found in '/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/PROC_sm_12/run_01_tag_1_debug.log'.
> Please attach this file to your report.
> ---------------------------------------------------------------------------------------------------
> The lhapdf and python interface is also there. I also untar a particular pdf set in the location "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/HEPTools/lhapdf6/share/LHAPDF/" . Still, this is showing. Do you have any idea what's wrong? Below there is a first few lines of the DEBUG file:
> ---------------------------------------------------------------------------------------------------
> generate_events run_01
> Traceback (most recent call last):
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/interface/extended_cmd.py", line 1501, in onecmd
> return self.onecmd_orig(line, **opt)
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/interface/extended_cmd.py", line 1450, in onecmd_orig
> return func(arg, **opt)
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/interface/madevent_interface.py", line 2469, in do_generate_events
> self.run_generate_events(switch_mode, args)
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/interface/common_run_interface.py", line 6711, in new_fct
> original_fct(obj, *args, **opts)
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/interface/madevent_interface.py", line 2551, in run_generate_events
> postcmd=False, printcmd=False)
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/interface/extended_cmd.py", line 1528, in exec_cmd
> stop = Cmd.onecmd_orig(current_interface, line, **opt)
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/interface/extended_cmd.py", line 1450, in onecmd_orig
> return func(arg, **opt)
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/interface/common_run_interface.py", line 1835, in do_systematics
> out = self.do_systematics(line)
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/interface/common_run_interface.py", line 1787, in do_systematics
> result=result_file
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/various/systematics.py", line 1027, in call_systematics
> obj = Systematics(input, output, log=log, **opts)
> File "/home/saumyen/HEP_Package/MG5_aMC_v2_6_1/madgraph/various/systematics.py", line 195, in __init__
> pdfset = lhapdf.mkPDF(int(data)).set()
> File "lhapdf.pyx", line 509, in lhapdf.mkPDF (lhapdf.cpp:8392)
> File "lhapdf.pyx", line 493, in lhapdf.mkPDF_lhaid (lhapdf.cpp:8188)
> RuntimeError: Info file not found for PDF set 'NNPDF23_lo_as_0130_qed'
> Run Options
> -----------
> stdout_level : 20 (user set)
>
> MadEvent Options
> ----------------
> automatic_html_opening : False (user set)
> notification_center : True
> cluster_temp_path : None
> cluster_memory : None (user set)
> cluster_size : 100
> cluster_queue : None
> nb_core : 4 (user set)
> cluster_time : 4 (user set)
> run_mode : 2
>
> Configuration Options
> ---------------------
> text_editor : None
> cluster_local_path : None
> cluster_status_update : (600, 30)
> pythia8_path : /home/saumyen/HEP_Package/MG5_aMC_v2_6_1/HEPTools/pythia8 (user set)
> hwpp_path : None (user set)
> pythia-pgs_path : None (user set)
> td_path : None (user set)
> delphes_path : /home/saumyen/HEP_Package/MG5_aMC_v2_6_1/Delphes (user set)
> thepeg_path : None (user set)
> cluster_type : condor
> madanalysis5_path : /home/saumyen/HEP_Package/MG5_aMC_v2_6_1/HEPTools/madanalysis5/madanalysis5 (user set)
> cluster_nb_retry : 1
> eps_viewer : None
> web_browser : None
> syscalc_path : /home/saumyen/HEP_Package/MG5_aMC_v2_6_1/SysCalc (user set)
> madanalysis_path : None (user set)
> lhapdf : /home/saumyen/HEP_Package/MG5_aMC_v2_6_1/HEPTools/lhapdf6/bin/lhapdf-config (user set)
> f2py_compiler : None
> hepmc_path : /home/saumyen/HEP_Package/MG5_aMC_v2_6_1/HEPTools/hepmc (user set)
> cluster_retry_wait : 300
> fortran_compiler : None
> auto_update : 7 (user set)
> exrootanalysis_path : /home/saumyen/HEP_Package/MG5_aMC_v2_6_1/ExRootAnalysis (user set)
> timeout : 60
> cpp_compiler : None
>
> ------------------------------------------------------------------------------------------------------------------------------------
> I think there is a file missing in that directory. But did that happen and how can I plug this?
> I will be very if this gets resolved.
>
> Regards,
>
> Saumyen
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hi Olivier,

>I guess that you are using lhapdf6.1.6 and that your PDF set is not NATIVELY supported by that version of lhapdf. This is the reason why the python link of lhapdf fails to link to that pdf set.

I was actually initially using a pdf set for LHAPDF6.2.1. But that I rectified and used the pdf sets for LHAPDF6.1.6only. But it is showing the same thing.

>You might check with lhapdf author, but I guess that it exists a lhapdf command like
>lhapdf-config update
>exists to update the database and to handle that pdf.

I don't actually know if there's any command like this that I have to give to give. I will try checking that with the lhapdf authors.

Thank you for your kind response.
And yeah, sorry for those typos in the previous message.

Regards,

Saumyen

Hi Olivier,

I talked to the LHAPDF authors, but they are saying that it may be MG5's concern. They are saying that MG5 try to check the LHAPDF data files by itself (though this is really LHAPDF's responsibility). So, it may have happened that LHAPDF installation is all right but somehow MG5 cannot find the datafiles or cannot find where to check.

Another thing is that whenever I'm installing LHAPDF with the "install lhapdf6" command, it is downloading and installing the lhapdf6 version 6.1.6 whereas the version 6.2.1 is there for almost a year or so. But usually, MG5 downloads the latest possible version, right? So why is it like this?

Another thing is that I was talking with one of my colleagues on this regard. He has done exactly the same as what I have. And it's running smoothly on his computer. Then what's wrong with me? He has also Ubuntu18.04 like me.

For your information, I am using the following compilers and programs: gcc version 7.3.0, Python version 2.7.15rc1, gfortran version 7.3.0, python-lhapdf version 5.9.1-6, Boost version 1.59 (automatically downloaded when installing lhapdf6 for the first time), ROOT 6.14/04 .

Any help in this regard will be much appreciated.
Regards,

Saumyen

Dear Saumyen,

> I talked to the LHAPDF authors, but they are saying that it may be MG5's concern.

Maybe you should add me in your contact with them and forward me their answer such that we have a place where we can all discuss. You can also ask them to join this thread if you prefer. (In the first case, you can attach my response below)

> They are saying that MG5 try to check the LHAPDF data files by itself (though this is really LHAPDF's responsibility). So, it may have happened that LHAPDF installation is all right but somehow MG5 cannot find the datafiles or cannot find where to check.

I would no say that we never check the datafile because we do (even if i agree that it should not be our responsibility), the reason for us doing it is that we want to have auto-download of the pdfset. Something that the lhapdf author does not want to support.

This point being clarified, the crash is not connected at all to the above point. In this case, the crash occur when I run the command:
pdfset = lhapdf.mkPDF(int(data)).set()
So I'm just calling one of the python API of lhapdf. And this is that one crashing... So I continue to claim that this is a lhapdf side issue.

> Another thing is that whenever I'm installing LHAPDF with the "install lhapdf6" command, it is downloading and installing the lhapdf6 version 6.1.6 whereas the version 6.2.1 is there for almost a year or so. But usually, MG5 downloads the latest possible version, right? So why is it like this?

When lhapdf 6.2.0 was released we realize that such version of lhapdf
1) was not using the same dependency as 6.1.6 so our "install lhapdf" has to be fully re-written if we want to support lhapdf6.2.x
2) that lhapdf6.2 can not be linked to our code in the same way as were interfacing lhapdf6.1.6.

The last problem has been fixed and the latest version of MG5aMC detects which version of lhapdf you are using and link to it in the appropriate way. Now we can not change the "install lhapdf" command to link to lhapdf 6.2 due to the way that command works.
If we do change that command, old version of MG5aMC not supporting lhpadf6.2 would also try to install it and this will lead to crash. In such situation we have to stick to the old version for ~2 years.

> Another thing is that I was talking with one of my colleagues on this regard. He has done exactly the same as what I have. And it's running smoothly on his computer. Then what's wrong with me? He has also Ubuntu18.04 like me.

Like this, I have actually no idea.
I still believe that this is an error on the lhapdf side (or maybe I'm not suppose to use the above command)
Now you try with lhapdf6.2 since the lhapdf author will not (and should not) provide support for lhapdf6.1.6 anymore.

> python-lhapdf version 5.9.1-6

What is that version number? This sounds weird to me. How did you get that code/version number?

Cheers,

Olivier

Hi Olivier,

Sorry for late response. Should I really add you to the email thread? I mean you are not offended, right?
The problem is still there and in Ubuntu 16.04 machine also. And due to this neither Pythia, nor Delphes, nor MadAnalysis is running. So, I can't literally do anything. Please let me know.
Thanks a lot again.

Saumyen

Hi,

> Should I really add you to the email thread?

This is up to you.

> I mean you are not offended, right?

Don't worry I'm not.

> The problem is still there and in Ubuntu 16.04 machine also. And due to this neither Pythia, nor Delphes, nor MadAnalysis is running. So, I can't literally do anything. Please let me know.

We have identify two"new" problems with lhapdf.
- lhpadf does not compile within Mojave (at least not within the MG5aMC framework), we have added dedicated flag at the compilation time to use old version of the compiler within clang to fix that issue. We notify the lhapdf author.
- the command "lhapdf install XXX" is not working in lhapdf (in any version of lhapdf) due to a change within hepforge website. We also reported the problem to the authors of lhapdf, and my understanding is that they will not fix it.
So wihtin MG5aMC, I have added a work around that if the official method to download the pdf set does not work then I'm letting MG5aMC doing it manually (and interact with hepforge myself).

If your problem was related in a way or another to the second issue, you might want to try MG5aMC 2.6.4. If this is not the issue then you need to contact lhapdf author.
Another option is to de-activated the systematics computation by setting use_syst to False in the run_card

Cheers,

Olivier

Hi Olivier,

This is now working fine after I installed Anaconda with Python3.7 in my Ubuntu18.04. I'm not sure if the problem was for any missing package or it was for the Python version, because in my system Python3.6 was already there. So, I'm not sure what solved my problem. Again, shockingly I just found out that it may not be for Anaconda as I downloaded it in another system of mine with Ubuntu16.04 but it's still not working. Does root play any role here. Because in the Ubuntu16.04 system root isn't there.
Can you help me on this confusion?
Best regards,

Saumyen

Root is not needed for lhapdf.

MG5aMC does not support Python3.
It might just go trough since you do not have the linking of lhapdf trough python and therefore do not compute systematics.

Cheers,

Olivier

Hi Olivier,

No, It's calculating the systematics in the Ubutu18.04 system that I can see, because I have Python2.7 also. There's no problem in my Ubyntu18.04 system. But I have no idea it got fixed. May that will give me the answer as why it's not still working in my Ubuntu16.04 system.
Regards,

Saumyen

For Ubuntu 16.04,
I have followed the following recipe to make it work (but with lhapdf6.1.6)
https://www.singularity-hub.org/containers/5631

To have lhapdf 6.2.1 to compile on Ubuntu16,
I had to run the following command:
sudo apt-get install libboost-all-dev

otherwise it crashes with:
     g++ -O -O3 -I/usr/local/include -c -o pdf_lhapdf6.o pdf_lhapdf6.cc
     pdf_lhapdf6.cc:14:32: fatal error: boost/shared_ptr.hpp: No such file or directory
     compilation terminated.

But after that, I have it running correctly.

Cheers,

Olivier

> On 4 Dec 2018, at 15:22, SAUMYEN KUNDU <email address hidden> wrote:
>
> Question #675492 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/675492
>
> SAUMYEN KUNDU posted a new comment:
> Hi Olivier,
>
> No, It's calculating the systematics in the Ubutu18.04 system that I can see, because I have Python2.7 also. There's no problem in my Ubyntu18.04 system. But I have no idea it got fixed. May that will give me the answer as why it's not still working in my Ubuntu16.04 system.
> Regards,
>
> Saumyen
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.