[lammps-users] AIREBO not working in parallel

Hello,

I have a fluid layer confined between two walls. The fluid layer is made of hydrocarbons (Squalane). The wall atoms are made of gold atoms. The potentials for the fluid atoms are AIREBO. The potentials for walls are lj/class2 (no bonds, dihedrals etc…The walls are being fixed with ‘setforce 0 0 0’.) The entire system is well equilibrated (I mean none of the fluid atoms are nonphysically close in the system so as to be blown apart). I am using metal units, with a timestep of 0.0001 ps.

Now this the problem I am facing. I can run the simulation in one processor (serially). I have run for 100000 timesteps, which took around 3 days. The results are perfectly are as expected.
However when I submit the job in parallel, the system doesn’t run at all. The submitted job just stays (or hangs). There is no output. At the same time, the simulation doesn’t quit with any type of error.

I have run many other jobs in parallel with other potentials like PCFF and lj/cut. I have even run a Carbon Nano tube simulation with AIREBO potential in parallel. I am just not understanding why this particular configuration (with hydrocarbons) doesn’t run in parallel.

I am attaching my lammps input file and the corresponding read_data file. I am also attaching the ‘CH.airebo’ file.

I have to run simulation for at least 1 ns, which would take more than a month if run in serial. Any help is greatly appreciated.

Thank you

With Regards
Ajay

in.flow (1.52 KB)

SQ_Au.datafile (1.65 MB)

CH.airebo (770 KB)

hello ajay,

your input works fine for me, serial and in parallel.

what version of lammps are you using?
and what kind of hardware are you running on?

does it stall in parallel at all, regardless of how
many processors you use and how you address
the individual nodes?

axel.