[ESPResSo-users] Re: [ESPResSo] deleting bonds

[Axel Arnold]

On Thursday 09 December 2010 00:17:00 you wrote:
> Hi,
>
> it might be not very related to a problem but i had a question about
> p->l.i[0,1,2] which is supposed to be an index of the simulation box. Is it
> just 0,0,0 for the main simulation box and if the particle outside it just
> gives on what side of the main box it is ? SO like if it's 1,0,0 it would
> mean that the particle is outside to the right of the x axis ?

Almost. The positions are only folded up to the value of the skin global 
parameter. That is, if skin is 0.3, and boxlength 10, then a particle can have 
positions from -0.3 to 10.3. Therefore, a particle at -0.1 can either be 
stored with position -0.1 and i=0 or 9.9 and i=1. That is, the absolute 
position is always continuous.

> While i was
> trying to find an error I just noticed that the distance between two
> particles is sometimes really big( of the size of simulation box) and after
> I check their indexes p->l.i they appear to be different. I just thought
> before that since I check them on being the ghost they can't have such a
> large distance between them.

Well, depends on where the particles come from. If they are form neighboring 
cells, then no, their distances should be smaller than the cell size, but if 
you use bonds, they can point to particles at the other end of the main 
simulation box, with a corresponding large distance, which in fact is 
incorrect. So for bonded particles, you should use the get_mi_vector function 
for the minimum image distance.

Axel

-- 
JP Dr. Axel Arnold Tel: +49 711 685 67609
ICP, Universität Stuttgart      Email: address@hidden
Pfaffenwaldring 27
70569 Stuttgart, Germany