[gmx-developers] Computing the density of water around water

[Mark Abraham]

> Hi,
>
>     I know that one can get g2(r) (radial distribution) information from a
> simulation of identical particles by looping over all particles i and
> making a nested loop over all particles j < i.  Thus, one can compute
> (1/2)N^2 interactions from each snapshot for computing the histograms that
> go into each g2(r).  However, I'm not sure if one can do this for the
> three-dimensional density calculations.
>     If I simply want to plot the three-dimensional density of water oxygen
> atoms around a particular water molecule, is it OK to do a simple
> simulation of water (no constraints, no particle fixed at the center)

Of course - you don't need a fixed particle from which to measure. The
distribution of atom-atom distances is necessarily independent of the
reference frame in which they are measured. After the fact, you could move
your box to center it on a particular water at each time step, or you
could be perverse and have a box that moved around randomly and still all
the atoms have the same distribution of distances.

> and
> then analyze each snapshot by putting every water molecule i at the center
> and considering the oxygen atom of every other water j < i in my density
> histogram?  This would increase my sampling by N/2 times, and in my
> simulations N is 1728!  It seems like that should be all right, but I just
> wanted to check.

Sounds fine to me.

Mark