Replicate command

I have a data file for a lipid bilayer system in water (attached). It
runs fine as it is on the 10 April version of Lammps and the 28 Oct
version of Lammps.

I now want to replicate it in the x and y directions (ie to make a
bigger bilayer). The replicate command seems to do this in the 10
April version without error but if I try using replicate 2 2 1 in the
28 Oct version, I get the error "Bond/angle/dihedral extent > half of
periodic box length (domain.cpp:590)". I get this error whenever I
try and replicate in either the x or y directions; replicate in the z
direction works fine.

I don't understand why I'm getting this message as, according to my
reading of the manual, it means that the maximum extent of 2 bonds (I
have no dihedrals) is more than half the box side - but having
replicated, the box side is now twice the size it was when it worked.

I had a look back through this mailing list and it looks like a
similar question was asked on 18 September 2012 - but in that case the
error seemed to arise from restart files that might have been storing
the image flags in different ways depending on the version of Lammps.
I have no image flag data in data file I'm inputting. I've run the
system without any replication and dumped out the image flags for each
coarse-grained bead; all beads are initially in the 0,0,0 image.

Can anyone make any suggestions as to what might be going wrong and
how to fix it - there are features in the later version of Lammps I
need to use but I also need to reliably replicate the bilayer?



data.1024lip9216wat (3.32 MB)

forcefield.elba (2.25 KB)

in.elba (720 Bytes)

You're replicating a data file. Do you have any bonds
that cross periodic boundaries in the data file? If so,
you need to have image flags defined in order to replicate.


Sorry, I don't understand. As far as I can tell, there are no bonds
that cross periodic boundaries; I checked this by running the data
file as is (no replication) and using the dump command to tell me
which image each bead is in at the start of the simulation. They're
all in the same image (ix=iy=iz=0) so I assumed that meant that no
bonds crossed a boundary.


No - it doesn't mean that. They are all 0 b/c your data
file didn't specify them. Incorrect image flags
don't really affect things unless the atoms get
unwrapped by a command, which is what the replicate command
is doing. I need to think if there is a simple way to
keep the replicate command or the error test from getting confused.