From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Mon Feb 13 2012 - 16:21:16 CST

On Mon, Feb 13, 2012 at 4:21 PM, Paymon Pirzadeh <ppirzade_at_ucalgary.ca> wrote:
> Hello,
> I am trying to calculate the radial distribution of water molecules with
> respect to a few surface residues of my protein. The final profile
> clearly is not properly normalized since apparently the module does not
> care about the empty volume portion properly. The profile starts from

there is no "proper" way.

> some values close to zero, with a peak at 1.75A (consistent with
> literature) and increases up to some number but not 1 in long distances.

a radial distribution function can only be normalized properly,
if you know the volume correctly. with periodic boundaries
that is easy; without, this is an ill defined property.
what is the volume of a single atom, please?

also, the limit of 1 is only achieved for a homogeneous
distribution of particles. if you'd do a g(r) of supercritical
water not too far away from the tripelpoint, for example,
it would be lower than 1.0 for long distances, since you
have density fluctuations, and thus it is more probable
to have more particles close by and less particles far away.

> I have removed pbc and I am updating the selection (along the
> trajectory). Can anyone tell me how I can properly normalize this
> function?

you cannot. see above. for non-pbc, any normalization is
arbitrary and it *has* to go to 0 for large r, since there are
no particles at large r.

axel.

> Best,
>
> Paymon
>

-- 
Dr. Axel Kohlmeyer
akohlmey_at_gmail.com  http://goo.gl/1wk0
College of Science and Technology
Temple University, Philadelphia PA, USA.