Inconsistent image flags - regd

Dear all,

I am using rerun command to compute pe/atom. In my original run I have used triclinic box with constraint box tilt large (as box tilt is large) and moreover I didn’t allow the box to flip during the fix npt, though inefficient (as I am interested in time evolution of lattice parameters during NPT run) and wrote unwrapped coordinates in DCD format using dump dcd.

Now I am performing rerun on this trajectory by using the following input file

units real
dimension 3
boundary p p p
atom_style full
box tilt large
log name.log
neighbor 2.0 bin
neigh_modify every 1 delay 5 check yes
read_restart name.restart
include parm.XL
group polyA1 id 4936
compute 1 polyA1 pe/atom
compute pe1 all reduce sum c_1
thermo 1
thermo_style custom step c_pe1
rerun md_300k.dcd first 0 last 15000000 every 0 dump x y z box yes scaled no format molfile dcd .:/home/vmd/vmd-1.9.1/plugins/LINUXAMD64/molfile

with this script I am getting the warning “WARNING: Inconsistent image flags (…/domain.cpp:600)”.
As a checking I have calculated total PE of system using “compute pe” and compared with original run total PE, which is not exactly the same but with more or less with 2 kcal/mol difference in each step. More over I am using very latest version of lammps also.

Can any body please comment on this.

Thank you in advance.

Best,

Ramesh

Dear all,

           I am using rerun command to compute pe/atom. In my original run I
have used triclinic box with constraint box tilt large (as box tilt is
large) and moreover I didn't allow the box to flip during the fix npt,
though inefficient (as I am interested in time evolution of lattice
parameters during NPT run) and wrote unwrapped coordinates in DCD format
using dump dcd.

Now I am performing rerun on this trajectory by using the following input
file

units real
dimension 3
boundary p p p
atom_style full
box tilt large
log name.log
neighbor 2.0 bin
neigh_modify every 1 delay 5 check yes
read_restart name.restart
include parm.XL
group polyA1 id 4936
compute 1 polyA1 pe/atom
compute pe1 all reduce sum c_1
thermo 1
thermo_style custom step c_pe1
rerun md_300k.dcd first 0 last 15000000 every 0 dump x y z box
yes scaled no format molfile dcd
.:/home/vmd/vmd-1.9.1/plugins/LINUXAMD64/molfile

with this script I am getting the warning "WARNING: Inconsistent image
flags (../domain.cpp:600)".

which command of the script exactly produces this warning?

As a checking I have calculated total PE of system using "compute pe" and
compared with original run total PE, which is not exactly the same but with
more or less with 2 kcal/mol difference in each step. More over I am using
very latest version of lammps also.

Can any body please comment on this.

hard to make any specific comments. dcd stores coordinates in single
precision floating point, so you lose about 7 digits precision
relative to the double precision floating point that LAMMPS uses
internally, and that truncation gets even larger, if your positions
wrap around periodic boundaries. whether this is sufficient to explain
the difference you see, depends on the size of your system and the
average potential/kinetic energy. you could make a sensitivity
analysis using the displace atoms command where you randomize atom
positions at the same order of magnitude and see for yourself.

other than that, there could be an issue related to the image flag
warning, which can have several causes (some problematic, some not).

Dear Axel,

Thank you very much for your reply. Actually I am getting the warning with the following line of the input script

rerun md_300k.dcd first 0 last 15000000 every 0 dump x y z box yes scaled no format molfile dcd .:/home/vmd/vmd-1.9.1/plugins/LINUXAMD64/molfile

Thank you,

Best,

Ramesh

Dear Axel,

  Thank you very much for your reply. Actually I am getting the warning
with the following line of the input script

rerun md_300k.dcd first 0 last 15000000 every 0 dump x y z box
yes scaled no format molfile dcd
.:/home/vmd/vmd-1.9.1/plugins/LINUXAMD64/molfile

i don't see any obvious issues in the current code related to this and
i don't have an easy way to reproduce what you are doing, so you are
essentially on your own to debug this. however, from your quote of the
warning, it looks like you are using a significantly outdated version
of LAMMPS, so you may want to update to the latest version that
includes all bugfixes and improvements first and then re-check and
start debugging your problem.

axel.

Dear Axel,

Thank you very much for your reply, Actually I am using very recent version of LAMMPS (i.e., 30-oct-2014), with this version also I am getting the same WARNING.

Is this might be due to skewed triclinic box as I am bypassing this by using “box tilt large” in my original run as well as in rerun (otherwise job is killing with the error ERROR: Triclinic box skew is too large).

Thank you,

Best,

Ramesh

Dear Axel,

  Thank you very much for your reply, Actually I am using very recent
version of LAMMPS (i.e., 30-oct-2014), with this version also I am getting
the same WARNING.
Is this might be due to skewed triclinic box as I am bypassing this by using
"box tilt large" in my original run as well as in rerun (otherwise job is
killing with the error ERROR: Triclinic box skew is too large).

with using this option, you are asking for trouble. again, this falls
into the SEP category and you will have to check and debug yourself if
LAMMPS does the right thing. most of the code in LAMMPS that handles
non-orthogonal boxes was written before this option was available and
there is no way to easily determine which parts make assumptions about
the tilt factors being well behaved or not.