lattice fcc 4.08
region box block 0 5 0 3 0 5
boundary p f p
create_box 1 box
create_atoms 1
mass 1 196.9665
velocity all create 293 234325
velocity all scale 293
pair_style eam
pair_coeff * * auu3
fix 1 all nve
fix 2 all temp/rescale 293 293 1 10 0.5
fix 3 all msd 1 all.out
restart 100 r1 r2
timestep 0.005
thermo 50
run 1000
The restart file r2 is created so I use it for the restart run:
units metal
atom_style eam
read_restart r2
pair_style eam
pair_coeff * * auu3
fix 1 all nve
fix 2 all temp/rescale 293 293 1 10 0.5
fix 3 all msd 1 all.out
timestep 0.005
thermo 50
run 1000
I just ran it and got the error: Did not assign all atoms correctly. Also a
warning before the error: Resetting boundary settings to restart file
values.
The problem with restarts in your boundary command in the 1st run. It
is non-perioidic in y, yet you have atoms that cross the boundary. They
weren't lost b/c you were probably only running on a single processor. But
when that configuration is written to a restart file and then read-in they
will be considered out of the box and lost. You should not use a fixed
non-periodic boundary unless you want atoms to be lost or you insure
atoms do not cross it.
The reason why I'm considering y fixed is because I want to represent a
surface. So, how could I write a restart file for this case? I'm doing the
first run (to create the restart file) for minimization purposes.
Put the boundary far enough away from the surface that atoms
will never cross it. Or use a shrink-wrap boundary (s) that will
conform to the surface automatically.