[gmx-developers] pbc.c: put_atoms_in_triclinic_box

[Berk Hess]

>I would like to make a suggestion for replacing the function
>put_atoms_in_triclinic_box in pbc.c, because the current one, which I
>find a bit hard to understand, produces a shift between the box and the
>atoms. The following puts all of the atoms in the triclinic box:

Whether or not it produces a shift depends on definitions.
Currently all put_atom_in_?_box routines keep the center of the box at
the center. And the center is defined as a half times the diagonal of the 
box matrix.
So this can be changed easily and consistenly by changing the center 
definition.

>void put_atoms_in_triclinic_unitcell(matrix box, int natoms, rvec x[])
>{
>   matrix S,T;
>   int    i,j;
>
>   transpose(box, S);
>   m_inv(S,T);
>   for(i=0; i<natoms; i++)
>   {
>     mvmul(T, x[i], x[i]);
>     x[i][0] = trunc(x[i][0]);
>     x[i][1] = trunc(x[i][1]);
>     x[i][2] = trunc(x[i][2]);
>     mvmul(S, x[i], x[i]);
>   }
>}

This will not work. You don't want trunc, you want the remainder.

But again, the routine should not be changed, only the calc_box_center 
routine.
We can discuss which definition is better/more useful.
In rasmol there seems to be a shift, but other programs might work 
differently.

Berk.

_________________________________________________________________
Talk with your online friends with MSN Messenger http://messenger.msn.nl/