[lammps-users] Lost atoms in pbc simulation!

Hello everyone:
A perfect solid structure is simulated in lammps.The period boundary condition is used. When the temperature is low, the program runs normally. When the temperature is so high, for example,10000k, some atoms are lost.But in my opinion, the atom should not be lost in period boundary conditon. Who can give me some explations for this? Thanks! |

Hello everyone:
     A perfect solid structure is simulated in lammps.The period boundary condition is used. When the temperature is low, the program runs normally. When the temperature is so high, for example,10000k, some atoms are lost.But in my opinion, the atom should not be lost in period boundary conditon. Who can give me some explations for this? Thanks!

this means, that your input script is not suitable for these conditions. keep
in mind that at high temperatures (some) atoms move _much_ faster.
in LAMMPS atoms are distributed between domains and they get lost
when atoms move too fast for the neighbor list rebuild to be able to
communicate them between neighboring domains.

please read the LAMMPS documentation and LAMMPS paper to get more
details about what "lost atoms" mean and what to do about it.

axel.

If your dynamics is bogus, e.g highly overlapping atoms shooting long distances,
then you can lose atoms.

Steve