fix move problem after lost atoms

Hi everyone, I am using Lammps to simulate impinging nanoscale jets, and the fix move command hehaved badly after some atoms moved across the non-period boundary and lost.
The fix move command was applied on a solid wall composed of gold atoms making it moved in constant veclocity pushing some argon atoms to emit from a slit. And the problem is that after some argon atoms was lost the veclocity of the wall is much faster than what it was set by the fix move command. I set the boundary type to be ‘p f p’, and toggled the ‘lost’ option of the thermo_modify command to be ‘warn’. The fix move command had behaved quite well before the warning of lost atom, but not after the warning. Furthermore after some timesteps I found through ‘dump custom’ command about half of the argon atoms was lost while in VMD displayer I could see only very small fraction of argon atoms was lost.
Is anyone has any clue about this problem? I am using lammps-11Mar11. The simulation was run on 20 processor.
If I hadn’t provided enough information, pleased let me know. Much thanks.
PS. This is the second time I sent this message, I canceled the last one after I found a serious typo, apologize if this troubles you.

I don’t know what you mean by:

the fix move command hehaved badly

If you use fix move to move atoms outside a fixed boundary,
the atoms will be lost. If you don’t want this behaviour then
use shrink-wrap boundaries.

I also suggest you try the current version of LAMMPS.