AMBER Archive (2009)

Subject: [AMBER] Parallelepiped periodic box

From: Ian Streeter (
Date: Mon Jan 12 2009 - 09:57:55 CST

Dear Amber users,

I am trying to use Sander for molecular dynamics using a periodic
boundary condition. I want to use a periodic box that is a
parallelepiped but not rectangular (none of the angles are 90 degrees).
I have a couple of questions relating to how the periodic box is defined
in the Sander input and output files. I have not been able to find the
answers to these questions in the Amber manual or in the faq or in the
mailing list archive.

First question:

In the inpcrd file and rst files, the coordinates of the atoms are given
using an orthogonal coordinate system, and then the periodic box lengths
and angles are given at the end of the file. What is the relationship
between these lengths/angles and the vectors that define the periodic
box in the orthogonal coordinate system? Is it the same relationship as
for the pdb file format? i.e. is it the same as the following, which I
got from the protein databank contents guide:

If vector a, vector b, vector c describe the crystallographic cell
edges, and vector A, vector B, vector C are unit cell vectors in the
default orthogonal Angstroms system, then vector A, vector B, vector C
and vector a, vector b, vector c have the same origin; vector A is
parallel to vector a, vector B is parallel to vector C times vector A,
and vector C is parallel to vector a times vector b (i.e., vector c*).

Second question:

In the prmtop file, the periodic box is defined under the BOX_DIMENSIONS
flag. But it seems to be defined differently compared to the inpcrd
file - only one angle is specified, which the amber website says is "the
angle between the XY and YZ planes in degrees". No other angles may be
specified. Does this mean the box must be a 90 degree rectangle in the
XY plane?

Also, what is the exact meaning of the box lengths written in the prmtop
file? Are they the lengths of the edges of the parallelepiped, or are
they the coordinates of the box vertices using the orthogonal coordinate
system? Note these are not equivalent when the angle of the periodic
box is not 90 degrees.

If anyone thinks they can answer any of these questions I would be
extremely grateful.

Many thanks,


AMBER mailing list