[gmx-users] How to recenter solvent around solute

[Tsjerk Wassenaar]

Hi,

Justin is very right to point at the workflow. In your case, the
points 5 and 6 apply: centering, and putting things in some box. Maybe
that doesn't work in one pass and you'll have to use trjconv twice.

>> This does not work. Neither is the solute always surrounded by solvent for
>> every frame (it goes out of the box) nor is the original unit cell
>> (truncated octahedron) preserved.

You have to think out of the box with PBC :) The solute is always
surrounded by solvent, except where it touches its own image, in which
case you should consider throwing away and redo the run in a larger
box. There is no out of the box where atoms can go to. Furthermore,
the unit cell does not have shape, it only has translational relations
(lattice vectors). You can carve out any general shape you want, such
as rectangular, truncated octahedron, rhombic dodecahedron. Note that
only in special cases you can carve out regular instances of those
(such as a cube for a rectangular brick).
To understand this, here's a bit of an exercise:

Draw three identical hexagons in contact with each other
>From one of the hexagons, draw vectors to the other two, and draw the
parallellogram spanned by these vectors.
Place a hexagon at the farthest node, and not how the parallellogram
has exactly the same content as any of the hexagons.
Now turn the parallellogram into a rectangular brick by translating a
part of it, and note how the brick still has the same contents as a
hexagon.
Feeling confident that any shape will do, find some tilings from
Escher and try to see how you can fit that into a parallellogram,
rectangle, or hexagon :) Of course the coolest would be to use those
to define a unit cell to put your atoms in :p

Groetjes,

Tsjerk


-- 
Tsjerk A. Wassenaar, Ph.D.

post-doctoral researcher
Molecular Dynamics Group
* Groningen Institute for Biomolecular Research and Biotechnology
* Zernike Institute for Advanced Materials
University of Groningen
The Netherlands