Re: help with how pbc are defined.. .

From: David Hardy (
Date: Fri Oct 12 2012 - 12:16:14 CDT

Hi Erik,

The simulation is performed in the parallelepiped defined by the cell basis vectors. Given appropriately defined basis vectors, the octahedron shape that you obtain for your position output by using "wrapNearest" has equivalent volume and periodicity to the parallelepiped, i.e., they are equivalent space-filling shapes.

Best regards,

On Oct 12, 2012, at 9:39 AM, Martin, Erik W wrote:

> Thanks for the input. I'll add wrapNearest to the configurations file.
> I'm aware that its cosmetic, but I frequently have to give presentations
> to people not familiar with MD and I really prefer not answering questions
> about unwrapped trajectoriesŠ
> Maybe I wasn't entirely clear about the question though. What I wasn't
> 100% clear about was just how the boundary conditions are defined for
> these non-cubic systems. I was hoping someone could either give me a
> little information or point me to a good reference to read about this.
> Thanks again,
> Erik
> On 10/11/12 1:34 PM, "David Hardy" <> wrote:
>> Erik,
>> For your NAMD configuration file, using "wrapNearest" with "wrapWater"
>> should give your DCD trajectory the appearance of the octahedron without
>> you having to do any additional wrapping of the positions. I don't know
>> if the PBC wrapping in VMD has the option to apply a "wrapNearest" (i.e.,
>> wrap to the image closest to the center) to an unwrapped trajectory. You
>> probably already know that any kind of periodic wrapping is simply
>> cosmetic and doesn't actually affect the simulation.
>> -Dave
>> On Oct 11, 2012, at 12:48 PM, Martin, Erik W wrote:
>>> I've been using orthogonal boundary conditions for ages, and thought I
>>> had a good handle on how this worked. However, I recently had to start
>>> using a truncated octahedron and its become a bit of a mystery to me. I
>>> followed what instructions I could find and defined my basis vectors as:
>>> vector1 d 0 0
>>> vector2 -1/3d 2/3sqrt(2) d 0
>>> vector3 -1/3d -1/3sqrt(2)d -1/3sqrt(6)d
>>> Or specifically in my case:
>>> periodic cell dcdUnitCell yes cellBasisVector1 110.906 0.0 0.0
>>> cellBasisVector2 -36.9688 104.5635 0.0 cellBasisVector3 -36.9688
>>> -52.2818 -90.5547 cellOrigin 0.0 0.0 0.0
>>> The first structure I started appears to be running just fine.
>>> However, I left the line "wrapWater" out of the configuration file.
>>> When I set up a second structure (all I changed was the water model to
>>> tip4) I included this line. The simulation changed the structure to
>>> some manner of parallelogram (see attached picture tip4_box1
>>> Š The starting structure
>>> is tip4_starting I
>>> visualized the PBC with "pbc box" and saw that this was the shape of my
>>> boundary conditions ­ albeit shifted. There appeared to be no error in
>>> the running of the simulation, just this change. When I went back and
>>> looked at the tip3 simulation, I found that the shape of the boundary
>>> condition was the same, but seemed to have no impact on the structure
>>> (picture tip3_box1:
>>> However, when I used "pbc wrap" is forced everything into this boxŠ
>>> oddly enough, when I attempted to "unwrap" the tip 4 simulation it
>>> didn't return to the truncated octahedron.
>>> Is there something I'm missing here about how the boundary conditions
>>> are defined in the case of a truncated octahedron? Have I done
>>> something wrong? Perhaps its simply not possible to use wrapWater with
>>> this type of symmetry?
>>> Thanks a lot for any help,
>>> Erik
>>> ________________________________
>>> Email Disclaimer:
>>> Consultation Disclaimer:
>> --
>> David J. Hardy, Ph.D.
>> Theoretical and Computational Biophysics
>> Beckman Institute, University of Illinois

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:22:10 CST