[gmx-users] question about PBC and link cell based cutoff method

[Mark Abraham]

linfu wrote:
> Dear GMX community!
> 
> I have question about PBC and link cell based cutoff method.
> 
> (1)
> for example one water molecule,if only H atom of one water molecule
> move out of box, however, O atom are still in the box. does this H atom
> will need to be image back to the box through PBC,or just ignore it 
> until the mass center of one water move out of the box, then use PBC to 
> image back to the box. how does the gromacs do!
> 
> (2)
> I am confused about the link cell based cutoff method.i read some refs about
> this method,and it is very efficient.according to my understanding of precedure
> step1: divide the box into NxMxL cells.
> step2: put the all atoms into the cells
> step3: construct the neighboring cell index.
> If all the atoms do not move out of the whole box and do not apply PBC, i think 
> this method is quite efficient.
> Actually, we should apply the PBC for every md step, does this mean that we should
> update step2 and step3 every md step. if the answer is yes, why the cell based cutoff
> method is more efficient relative to sphere based cutoff.
> how does the gromacs implement.
> 
> Could anyone kindly help me to figure out these question.
> Thank you very much in advance!

It's hard to work out your problem, but it sounds like you haven't read
manual section 3.2 yet. Please do so, and get back to us.

Mark