[lammps-users] Questions concerning triclinic boxes.

Hello everyone,

I have the following lattice vectors:

a1 = 11.8281993900000000 0.0000000000000000 0.0000000000000000
a2 = -5.9140996950000000 10.2435211600000000 0.0000000000000000
a3 = 0.0000000000000000 0.0000000000000000 28.9730530900000000

(as some of you might guess, these come from a VASP POSCAR file)

My problem is turning this, and the positions of the atoms contained
in the DFT simulation, into a LAMMPS friendly format.

From the LAMMPS manual, a non-orthogonal simulation cell can be

defined using the following

a = (xhi-xlo,0,0)
b = (xy,yhi-ylo,0)
c = (xz,yz,zhi-zlo)

It would seem that the only tilt factor I need to consider is xy =
-5.914099695, with all the rest being set to zero. The values in the
lattice vectors make sense, since angle in the slab is 120 deg, which
is consistent with an fcc(111) lattice with a side that has been
aligned with the x-axis. The tilt factor I use, xy, would be the
distance that the upper-left vertex of the xy plane has to the x=0
line (the y-axis, i.e.), placing the origin (xlo, ylo, zlo) = (0, 0,

The box bounds printed in the *dump file are the following, at the top :

-5.9141 11.8282 -5.9141
0 10.2435 0
0 28.9731 0

This makes no sense to me. What is the meaning of these vectors?
This is important to me, because I rely on these vectors for some
unwrapping routines in my scripts.


They quantities are not vectors, they are a bounding box, followed
by the 3 tilt factors. From the dump doc page:

The simulation box bounds are
included in each snapshot and if the box is triclinic
(non-orthogonal), then the tilt factors are also printed; see the
"region prism"_region.html command for a description of tilt factors.
For triclinic boxes the box bounds themselves (first 2 quantities on
each line) are a true "bounding box" around the simulation domain,
which means they include the effect of any tilt.