water spc-model diffusion

Please keep the list in the loop.

If the temperature of the water is correct, then you
should get a reasonable difffusion coeff. Note that
to measure D you have to account for image flags,
as atoms move thru the periodic box,
which compute msd does, for example.

Also, SHAKE will affect a diffusion coeff, probably
in a small way, since a rigid body will likely diffuse
less easily than a flexible one.


Dear Steve,

I used commend:
compute msd all msd
fix msd all ave/time 100 1000 100000 c_msd[1] c_msd[2] c_msd[3] c_msd[4] file 222water1msd.txt
and I got the msd graph(8 ns), It's in the attached file.
It's obviously wrong.
I just change the box from (3nm,3nm,3nm) to (5nm,5nm,5nm), other factors are same.
In (3nm,3nm,3nm)box, density(0.8gcm-3), water can't diffuse. In (5nm,5nm,5nm)box, density(0.1gcm-3), water diffuses too fast.

BTW, Which software should I choose to get a water model, VMD, or MS?
Thank you for your attention.






I'm not sure what you're asking. LAMMPS is simply
measuring the MSD. If it's not "correct", then that
is an issue with how you setup your simulation.

There are many water models - I would find one
with a paper that says it does a good job with
diffusivity, and read carefully how they did the


BTW, Which software should I choose to get a water model, VMD, or MS?


For instance, this paper was submitted very recently to arxiv (see
also references therein):