Dear Users,
I find a bug in the pressure evaluation when running the attached script 'lammps.in' with the last version of lammps (5nov2016).
When using meam potential with cmin = 0.5 (in the Si_test.meam file), pressure and forces cannot be calculated (-nan) see in the log file.
But when I change cmin to 0.5001,
or replace one of the p boundary by s,
or change the lattice parameter from 5.431 to 5.42
or change the cell dimension from 3 lattice parameters to 2
the pressure is well calculated.
This bug was not present with the version named 16Fev16.
Both versions are compiled with the make_serial provided in the src directory, and MEAM librarys are compiled by gfortran compiler.
Do someone have an idea about the problem?
Thank you in advance
Sincerely
Julien
library.meam (417 Bytes)
Si_test.meam (19 Bytes)
lammps.in (400 Bytes)
log.lammps (2.43 KB)
Dear Users,
I find a bug in the pressure evaluation when running the attached script
'lammps.in' with the last version of lammps (5nov2016).
When using meam potential with cmin = 0.5 (in the Si_test.meam file),
pressure and forces cannot be calculated (-nan) see in the log file.
But when I change cmin to 0.5001,
or replace one of the p boundary by s,
or change the lattice parameter from 5.431 to 5.42
or change the cell dimension from 3 lattice parameters to 2
the pressure is well calculated.
This bug was not present with the version named 16Fev16.
Both versions are compiled with the make_serial provided in the src
directory, and MEAM librarys are compiled by gfortran compiler.
Do someone have an idea about the problem?
i cannot reproduce it on my machine. can you let us know which OS and
compiler versions you are using?
thanks,
axel.
Dear Users,
I find a bug in the pressure evaluation when running the attached script
'lammps.in' with the last version of lammps (5nov2016).
When using meam potential with cmin = 0.5 (in the Si_test.meam file),
pressure and forces cannot be calculated (-nan) see in the log file.
But when I change cmin to 0.5001,
or replace one of the p boundary by s,
or change the lattice parameter from 5.431 to 5.42
or change the cell dimension from 3 lattice parameters to 2
the pressure is well calculated.
This bug was not present with the version named 16Fev16.
Both versions are compiled with the make_serial provided in the src
directory, and MEAM librarys are compiled by gfortran compiler.
Do someone have an idea about the problem?
i cannot reproduce it on my machine. can you let us know which OS and
compiler versions you are using?
I am using :
OS: Ubuntu 14.04LTS, Linux 3.13.0-95-generic x86_64
MEAM : gfortran 4.8
Lammps : g++ 4.8
Thank you
Julien
Dear Users,
I find a bug in the pressure evaluation when running the attached script
'lammps.in' with the last version of lammps (5nov2016).
When using meam potential with cmin = 0.5 (in the Si_test.meam file),
pressure and forces cannot be calculated (-nan) see in the log file.
But when I change cmin to 0.5001,
or replace one of the p boundary by s,
or change the lattice parameter from 5.431 to 5.42
or change the cell dimension from 3 lattice parameters to 2
the pressure is well calculated.
This bug was not present with the version named 16Fev16.
Both versions are compiled with the make_serial provided in the src
directory, and MEAM librarys are compiled by gfortran compiler.
Do someone have an idea about the problem?
i cannot reproduce it on my machine. can you let us know which OS and
compiler versions you are using?
I am using :
OS: Ubuntu 14.04LTS, Linux 3.13.0-95-generic x86_64
MEAM : gfortran 4.8
Lammps : g++ 4.8
ok. i have gcc/gfortran 6.2 and 4.9 on my machine here.
can you try cleaning the lib/meam directory and then comment out the
F90FLAGS = -O
line with less aggressive optimization and see, if this resolves the issue.
after recompiling the meam library, deleting lmp_serial and
compiling/linking it again.
thanks,
axel.
Dear Axel,
Thank you for your fast answer. Now with less aggressive optimization option it works fine.
Thanks again.
Julien