[gmx-users] SPC/E water RDF

chris.neale at utoronto.ca chris.neale at utoronto.ca
Sun May 30 16:30:51 CEST 2010

Dear Francisco:

1. gromacs already has an spce.itp file. Try using that one directly  
and see if your answer changes. If it does, then one file is in error.

2. the density of spc/e is not 1g/cc. First thing, you should try to  
reproduce an RDF from the literature for which you use exactly the  
same conditions. You mention NPT so try it. Also, be sure that you are  
using the same cutoffs/PME/etc.

2b. If you can get the published RDF in part 2, then you can try  
varying one condition at a time, e.g. take your density-equilibrated  
NPT box and try running some NVT and ensure that you still get the  
correct RDF.


-- original message --

Dear users,

Since I'm new to MD, I coded SPC/E water in the NVT emsemble.

Basic ingredients:

(1) Number of water molecules N=216
(2)   Mass of O = 16
       Mass of H = 1
(3)   Box density (Periodic Boundary Condition) is exactly 1g/cc (L
=18.6205 Angstrom)
(4)   SPC/E LJ parameters from Berendsen's original 1987 paper
(5)   Coulomb interaction: Ewald sum
       Real space ewald: Minimum image convention
       Ewald convergence parameter for reciprocal sum = 5/L (L is the
box length in Ang)
       Fourier cut-off for ewald reciprocal sum |G| =2*pi*|n|/L, where |n| < 27
(6)   Nose-Hoover chain NVT (single chain, single thermostat per chain)
(7) Time step 1 fs, velocity verlet solver, RATTLE was used to
constrain bond lengths and angle, and correct velocities.
(8) 100 ps equlibration, 400 ps production (configuration was saved  
every 25 fs)

Everything is fine (energies are stable, etc.) except the second peak
in the O-O rdf.
The first peak is fine but my second peak seems flat. I find this
troubling since
all the rdfs I've seen in literature as well the experimental plot
have a well-defined
second peak. I must mention that in all the papers I've seen the
ensemble is NPT (I've not seen NVT yet). In any case, I was expecting  
my result
to match previously published results closely.

Can anyone point out what I am doing wrong?? Is NVT not a good ensemble
for water simulation?? I'll appreciate it if someone can refer me to a paper
that discusses this.

More information about the gromacs.org_gmx-users mailing list