MadGraph5_aMC@NLO

amcnlo compilation fail

Asked by Goutam Das

Dear developers,
   I was trying to do the following on a linux server :

p p > z z [QCD]
./bin/calculate_xsect

however it fails with the following error:
     mv libMadLoop.a ../
     make[1]: Leaving directory `/home/goutam/MADGRAPH/2.5.5/ZZ1/SubProcesses/P0_dxd_zz/V0_dxd_zz'
     gfortran -o check_poles parton_lum_1.o parton_lum_2.o parton_lum_3.o parton_lum_chooser.o matrix_1.o matrix_2.o matrix_3.o real_me_chooser.o fks_inc_chooser.o leshouche_inc_chooser.o genps_fks.o setcuts.o setscales.o born.o sborn_sf.o b_sf_001.o fks_Sij.o trapfpe.o fastjetfortran_madfks_full.o fastjet_wrapper.o fks_singular.o veto_xsec.o montecarlocounter.o reweight_xsec.o boostwdir2.o configs_and_props_inc_chooser.o initcluster.o cluster.o reweight.o get_color.o FKSParamReader.o iproc_map.o MC_integer.o reweight_xsec_events_pdf_dummy.o appl_interface_dummy.o cuts.o pythia_unlops.o recluster.o check_poles.o BinothLHA.o born_hel.o -L../../lib/ libMadLoop.a -lcts -liregi -L/home/goutam/MADGRAPH/2.5.5/HEPTools/lib/ -lninja -L/home/goutam/MADGRAPH/2.5.5/HEPTools/lib/ -lavh_olo -L../../lib/ -ldhelas -lgeneric -lmodel -lpdf -lcernlib -Wl,-rpath,/home/goutam/SOFTWARES/INSTALL/FASTJET321/lib -lm -L/home/goutam/SOFTWARES/INSTALL/FASTJET321/lib -lfastjettools -lfastjet -lfastjetplugins -lsiscone_spherical -lsiscone -lstdc++ -lstdc++
     /home/goutam/MADGRAPH/2.5.5/HEPTools/lib//libavh_olo.a(avh_olo.o): In function `__avh_olo_qp_print_MOD_printr':
     avh_olo.f90:(.text+0x87fcc): undefined reference to `_gfortran_transfer_real128_write'
     collect2: error: ld returned 1 exit status
     make: *** [check_poles] Error 1

Can you help me?

gcc --version
gcc (GCC) 5.2.1 20150902 (Red Hat 5.2.1-2)

gfortran --version
GNU Fortran (GCC) 5.2.1 20150902 (Red Hat 5.2.1-2)

Python 2.7.12

Thanks,
Goutam

Question information

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

This is a bug in some versions of libquadmath that is outside our control.
I have rewritten the parts of MadLoop that are sensitive to this bug, but not the version of OneLOop used by CutTools, which is in principle outside of our juridiction.
Now there are two ways to solve this:

a) Go through the OneLOop source code in HEPTools/oneloop and find all print statements of quadruple precision quantities, and rewrite them by first casting them into double precision.
That is, if you have:
"""
write(*,*) MY_QUAD_PREC_VARIABLE
"""
into
"""
MY_DOUBLE_PREC_VARIABLE = MY_QUAD_PREC_VARIABLE
write(*,*) MY_DOUBLE_PREC_VARIABLE
"""

b) Upgrade your gcc compiler version, in particular libquadmath. Apparently this bug is present in libquadmath shipped with gfortran 4.7 and was fixed thereafter (see https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51007)

Can you help with this problem?

Provide an answer of your own, or ask Goutam Das for more information if necessary.

To post a message you must log in.