Hello,
LAMMPS version I currently use is 17Nov2016.
I want to use a fixed boundary (style f) for a simulation, and also retain the particle-ID’s set at the first time step throughout the simulation.
One way to do this would be to use; neigh_modify once yes , but I am also instructing LAMMPS to write restart files and apply the fix_deposit command, which according to the manual will relist the neighbours - thereby changing the ID from the top.
Is there a way to retain the same particle-IDs, given the above conditions.
Could the boundary style m, be useful for this, but then how to ascertain which atoms crossed the boundary.
Regards
-Hari Tejas S Iyer
Hello,
LAMMPS version I currently use is 17Nov2016.
I want to use a fixed boundary (*style f*) for a simulation, and also
retain the particle-ID's set at the first time step throughout the
simulation.
One way to do this would be to use; *neigh_modify once yes* , but I am
also instructing LAMMPS to write restart files and apply the fix_deposit
command, which according to the manual will relist the neighbours - thereby
changing the ID from the top.
you are mixing up two completely different things. assignment of atom IDs
is completely unrelated to neighbor listing. not recomputing neighbor
lists is a very bad idea and makes no sense at all for a system when you
deposit atoms (don't you want those newly added atoms to interact? they
have to be in and have a neighbor list for that).
fix deposit gives you a choice for how it will determine what atom ID is
assigned to a new particle, but that choice has absolutely no impact on
neighbor list construction. an atom ID is a property the will stay an atom
for as long as it isn't removed from it (e.g. by passing through a fixed
boundary or explicitly deleted). the "id" keyword for fix deposit matters
only in this latter case.
Is there a way to retain the same particle-IDs, given the above conditions.
atoms do *not* change atom IDs unless you explicitly reassign them.
Could the *boundary style m*, be useful for this, but then how to
ascertain which atoms crossed the boundary.
this is another confused statement. with boundary "m", there is no
periodicity and the box is automatically extended, so atoms _by
construction_ cannot cross boundaries. however, this has absolutely nothing
to do with atom IDs.
if you believe that LAMMPS is not behaving in the way i have just
outlined (and what the manual really says), please provide a small(!) test
input deck to demonstrate it and explain how you determine that the code is
not matching what the documentation says.
axel.
Thanks, I understand now. just rechecked the output I was loosing the last atoms from the end of the initial list while using boundary style f. And I guess I misunderstood rebuilding of a neighbour list as a relisting where ids are reassigned, now I know that the ids are unique. I was worried that the atom ids might change if i loose atoms over a fixed boundary.
As for my question, I should use boundary style s/m to avoid particle deletion.
Would it be correct to say, that style m changes the box size and includes the atoms across the boundary in further interactions no matter how far they spread, and style s just shows spread atoms instead of deleting them as happens in style f.
Regards
Hari Tejas S Iyer
Thanks, I understand now. just rechecked the output I was loosing the last
atoms from the end of the initial list while using boundary *style f*.
And I guess I misunderstood rebuilding of a neighbour list as a relisting
where ids are reassigned, now I know that the ids are unique. I was worried
that the atom ids might change if i loose atoms over a fixed boundary.
As for my question, I should use boundary *style s/m* to avoid particle
deletion.
Would it be correct to say, that *style m* changes the box size and
includes the atoms across the boundary in further interactions no matter
how far they spread, and *style s* just shows spread atoms instead of
deleting them as happens in *style f*.
m and s are mostly the same. the only difference is that with m the box
will not shrink below its initial size.