From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Thu Mar 10 2011 - 09:44:13 CST

2011/3/10 Olaf Lenz <olenz_at_icp.uni-stuttgart.de>:
> Hi!
>
> On 03/10/2011 01:36 PM, Taco de Wolff wrote:
>> ITEM: BOX BOUNDS xy xz yz 0 6 0 0 6 3 0 6 0
>
>> Does _not_ show to be triclinic, eventhough by specification it
>> should (shouldn't it?).
>
> I do not know the lammpstrj format, but the above values look fishy.

olaf,

the box description in lammps is a bit unusual.
on top of that, they are boundaries and tilt "factors",
this all doesn't translate easily into vectors and matrices.
there is a bit of math needed to convert those.

this is a tilted box which boundaries are a cubic 6 angstrom cell,
but it is tilted by 3 angstrom in the y direction, i.e. has a
45 degrees beta angle.

the notation is xlo, xhi, x-tilt, ylo, yhi, ytilt, xlo, zhi, ztilt.

> Depending on whether they are denoted as column or row vectors, you
> either have two parallel axes (0 6 0 occurs twice), or one of them is
> the zero vector (0 0 0). None of this looks useful as unitcell vectors.
> So my guess is that the error handling of the file reader plugin is not
> perfect, as it does not complain, but that mainly your box vectors are
> wrong.

the issue is that the trajectory format was extended not so long ago to
cover this case and the box boundaries are still read correctly and can
be displayed, but of course that doesn't give you the correct unitcell.
there is an explanation that this notation would make it easier for
visualization programs, but i personally don't see where, and then again,
the simulation world is full of arbitrarily designed and strange ad hoc
file formats. ;)

cheers,
     axel.

> Olaf
>
> --
> Dr. rer. nat. Olaf Lenz
> Institut für Computerphysik, Pfaffenwaldring 27, D-70569 Stuttgart
> Phone: +49-711-685-63607
>

-- 
Dr. Axel Kohlmeyer
akohlmey_at_gmail.com  http://goo.gl/1wk0
Institute for Computational Molecular Science
Temple University, Philadelphia PA, USA.