Hi all,
During my simulation, there is always atom lost half way into the run, so i can never finish the run.
What are the possible causes for atom lost?
here is my fix:
timestep 0.003
thermo 5000
dump mydump all custom 3000 dump.multilayer id type x y z
fix 0 mobile npt temp 300 1750 2 iso 0 0 2
#fix 0 mobile npt temp 300 1750 2 iso 0 0 2
run 60000 every 1000 &
“group lreservoir region left” &
“set group lreservoir type 1” &
“group rreservoir region right” &
“set group rreservoir type 2”
unfix 0
fix 1 mobile npt temp 1750 1750 100 iso 0 0 2
run 600000 every 1000 &
“group lreservoir region left” &
“set group lreservoir type 1” &
“group rreservoir region right” &
“set group rreservoir type 2”
unfix 1
#fix 2 mobile npt 300 1000 2 xyz 0 0 2
fix 2 mobile nph iso 0 0 2
run 900000 every 1000 &
“group lreservoir region left” &
“set group lreservoir type 1” &
“group rreservoir region right” &
“set group rreservoir type 2”
unfix 2
write_restart fcc001sample
Hi all,
During my simulation, there is always atom lost half way into the run, so i
can never finish the run.
What are the possible causes for atom lost?
too many to name them all. usually it means that
you are doing something wrong. for more detailed
suggestions, have a look at the documentation and
search the list archives.
here is my fix:
timestep 0.003
thermo 5000
dump mydump all custom 3000 dump.multilayer id type x y z
fix 0 mobile npt temp 300 1750 2 iso 0 0 2
#fix 0 mobile npt temp 300 1750 2 iso 0 0 2
run 60000 every 1000 &
"group lreservoir region left" &
"set group lreservoir type 1" &
"group rreservoir region right" &
"set group rreservoir type 2"
unfix 0
hmmm... are you aware that in order to redefine a
group, you first have to delete it? otherwise, atoms
are only added to that group.
axel
Totally not aware of that. I just want to change the atom type to all atoms in that region, and i figure this is the only way to do it.
So how do i delete a group? Would that delete that actual atoms or just the group?
Basically i want to change atom type from 1 to 2 to the atoms in the specified group
Totally not aware of that. I just want to change the atom type to all atoms
in that region, and i figure this is the only way to do it.
So how do i delete a group? Would that delete that actual atoms or just the
group?
group delete
RTFM!
why should it delete atoms? it is a group.
axel.
Sorry i am still learning LAMMPS. i am an undergraduate and never had experience with this kind of tools before. And it is hard to read the documentation with limited knowledges in physics.
I appreciate your help very much:)
right now my main problem is the atom lost issue. and as u said there are too many possible causes to this issue.
I tried using different timesteps, and tried changing dampen units, but every LAMMPS simulation takes too long to run so i couldnt just throw a bunch of jobs at the computer and wait for the output.
And compiling LAMMPS on HHPC is a pain, semester is almost over but i could only get it to compile on the Mac workstation but not on school’s HHPC
Sorry i am still learning LAMMPS. i am an undergraduate and never had
experience with this kind of tools before. And it is hard to read the
documentation with limited knowledges in physics.
well, being able to read and understand documentation has
very little to do with physics, and very much with common sense.
it is ok to ask questions, but you have to show that you
are first making an effort to figure it out for yourself.
cheers,
axel.
I appreciate your help very much:)
right now my main problem is the atom lost issue. and as u said there are
too many possible causes to this issue.
I tried using different timesteps, and tried changing dampen units, but
every LAMMPS simulation takes too long to run so i couldnt just throw a
bunch of jobs at the computer and wait for the output.
if you run a smaller system, it will be _much_ faster!
i would wonder, if this strange changing of atom types has an
impact. i don't know how different the parameters between your
type 1 and type 2 are. if they are different, each reassignment of
atom types means that you push atoms out of equilibrium and
that can do very bad things to your simulation.
just run a single simulation that loses atoms quickly with
frequent output to a trajectory file and have a close look at
what happens briefly before you lose the atoms. there is
often some kind of unusual movement.
cheers,
axel.
My atom type 1 and 2 are Al and Ni.
They have different masses and i figured reassignment of atom type would probably do some weird stuff to the system. But since my simulation requires periodic boundaries, and i need to maintain the Al and Ni concentration close to 100% at the left and right edges, this seem to be the only way to do it.
But on a second thought, since my region originally started with pure Al atoms or Ni atoms, and im just replacing the whole region of atoms back to their original types, it shouldnt push out or destroy equilibrium.
anyway thanks so much for your help tonight, i will definitely read the manual before asking questions.
My atom type 1 and 2 are Al and Ni.
They have different masses and i figured reassignment of atom type would
probably do some weird stuff to the system. But since my simulation requires
it will! if you change a particles mass that has a velocity, then you
change its kinetic energy. remember: Ekin = 1/2 m v^2
periodic boundaries, and i need to maintain the Al and Ni concentration
close to 100% at the left and right edges, this seem to be the only way to
do it.
if you use periodic boundaries, there have to be atoms of the same kind on
both sides of the boundary. if you want to have a two-phase system, you
have to have at least two(!) phase boundaries. e.g. [1/2 Ni | 1/1
Al | 1/2 Ni]
you should visualize the system in a program that can display periodic
copies (e.g. VMD) and examine your system to make sure that what you
are giving as input to your simulation does make sense.
these are all quite fundamental issues, and thus it is highly recommended
to discuss this with your advisor. you are otherwise running a high risk of
wasting a lot of time doing calculations that are without any use.
choosing the right model is even more important than knowing how to
use an MD code well.
cheers,
axel.