lammps_spparks coupling example

Dear all,

I have build lammps_spparks which located in couple sub-directory.

But when I run the example like “mpirun -np 4 ./lmpspk 20 10.0 1 in.spparks”, I got a error “ERROR on proc 0: Invalid atom ID in Atoms section of data file (atom_vec_atomic.cpp:598)”.

It seems a lammps problem. We have tried this out many times, and the problem still apears.

Please help me if you kown the reason.

The related messages are list as follows:

$ mpirun -np 4 ./lmpspk 20 10.0 1 in.spparks

SPPARKS (20 Jan 2012)

LAMMPS (28 Mar 2012)

Created box = (0 0 -0.5) to (50 50 0.5)

2 by 2 by 1 processor grid

Creating sites …

2500 sites

2500 sites have 8 neighbors

Setting site values …

2500 settings made for site

Setting site values …

2500 settings made for d1

AT 0 1 0x853f860 1

AT 0 1 0x99b63d0 26

AT 0 1 0x875a3d0 1251

AT 0 1 0x857a208 43

AT 0 1 0x99f0d78 28

AT 0 1 0x8794d78 53

AT 1 3 0xb66d0008 0

AT 1 3 0xb66b4008 25

AT 1 3 0xb660a008 0

AT 1 3 0xb66d0010 0

AT 1 3 0xb66d0018 0

AT 1 3 0xb66b4010 0

AT 0 1 0x99a43d0 1276

AT 0 1 0x99ded78 67

AT 1 3 0xb665e008 25

AT 1 3 0xb660a010 25

AT 1 3 0xb66b4018 0

AT 1 3 0xb665e010 25

AT 1 3 0xb660a018 0

AT 1 3 0xb665e018 0

Reading data file …

orthogonal box = (0 0 -0.5) to (50 50 0.5)

2 by 2 by 1 MPI processor grid

ERROR on proc 0: Invalid atom ID in Atoms section of data file (atom_vec_atomic.cpp:598)

ERROR on proc 2: Invalid atom ID in Atoms section of data file (atom_vec_atomic.cpp:598)

application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0

Jianhua Li

03/30-2012

Dear all,

I have build lammps\_spparks which located in couple  sub\-directory\.

But when I run the example like "mpirun \-np 4 \./lmpspk 20 10\.0 1

in.spparks", I got a error "ERROR on proc 0: Invalid atom ID in Atoms
section of data file (atom_vec_atomic.cpp:598)".

It seems a lammps problem\. We have tried this out many times, and the

problem still apears.

actually, it is not a lammps problem, but a coupling problem, and
as such a result of the gradual switch to support larger systems
with 64-bit integer types in both SPPARKS and LAMMPS.

the coupling interface was apparently written before that change
and assumes 32-bit integers as atom identifiers, while by default
SPPARKS will now use 64-bit atom identifiers, which will lead
- due to the "glue code" in libcouple.a assuming them to be 32-bit
- to half of your atom ids in the generated LAMMPS data file
being set 0 (which is illegal).

to resolve this, is a non-trivial issue.

axle.