Hi Lammps users:
I just have two questions regarding the replicate command.
If ones starts with a restart file for an amorphous system and replicates the box, so the velocities are also replicated. After some MD run, say several million steps, is it safe to say that the “long range order” will be wiped out?
The other question is regarding the restriction of this command. On Lammps doc page, it says “You cannot use the replicate command to replicate a molecule that is bonded to itself across a periodic boundary”. I do have single bonds crossing the periodic boundary. However, since all chains are fairly short compared to the cell length, there is no chain striding through the entire cell. Can I use the replicate command?
Thanks.
Shaorui
Hi Lammps users:
I just have two questions regarding the replicate command.
If ones starts with a restart file for an amorphous system and replicates
the box, so the velocities are also replicated. After some MD run, say
several million steps, is it safe to say that the "long range order" will
be wiped out?
no. it is not at all safe to say. please think about it for a bit. the
only way to be safe is to measure it. please keep in mind that
decorrelation depends heavily on the properties of your material. on
the other hand, you can "aide" decorrelation by inserting randomness
into your system, e.g. by using fix langevin (preferably in the limit
of brownian dynamics) or regularly reassigning randomized velocities
or using dpd/tstat as an overlay. all of this will, of course, make it
rather pointless to replicate velocities.
The other question is regarding the restriction of this command. On Lammps
doc page, it says "You cannot use the replicate command to replicate a
molecule that is bonded to itself across a periodic boundary". I do have
single bonds crossing the periodic boundary. However, since all chains are
fairly short compared to the cell length, there is no chain striding through
the entire cell. Can I use the replicate command?
yes, you can, but it will only work correctly, if your initial
structure uses the image flags correctly and consistently.
axel.
Hi Axel
Thanks for replying. For the first question, you mentioned properties of materials. What kind of properties were you referring to? My system is full of very short chains (up to 6 beads per chain). Thus chains are quite easy to diffuse.
Also, regarding reassigning velocity, is it enough to reassign once right after replicating? Thanks.
Shaorui
Hi Axel
Thanks for replying. For the first question, you mentioned properties of
materials. What kind of properties were you referring to? My system is full
"everything". it is impossible to make a statement saying how many
steps are needed to decorrelate.
of very short chains (up to 6 beads per chain). Thus chains are quite easy
to diffuse.
doesn't matter. you have to *observe* or *measure*. there is no alternative.
Also, regarding reassigning velocity, is it enough to reassign once right
after replicating? Thanks.
same answer as above.
axel.