pressure vs density tip3p

From: Roy Kimura (
Date: Mon Aug 27 2007 - 15:54:40 CDT

Dear NAMD group,
I have been running boxes of pure water (216 TIP3P, starting box size ~
18.83 angstrom/side periodic box) in the NPT ensemble using the
Nose-Hoover Langevin piston method and temperature coupling (after 100
ps equilibration with velocity rescaling up to 298K). I am running at
1.0 atm pressure and 298 K but when I compute the water density during a
500 ps run by dividing the mass of the simulated water (216 * 18.016 /
6.022e23 grams) with the reported volume, I am getting an average value
of about 0.978 +/- 0.018 which is much less than the values reported in
the literature for TIP3P water (about 1.002, although experimentally it
should be about 0.997). I checked the time plots of all relevant values
(temperature, pressure, volume, density) and I am seeing no observable
drift, i.e., the system looks to be settled down correctly). The average
pressure value (-4.8 atm) was also reasonably close to the target
pressure (1.0) given the fluctuations of about +/- 840 atm. After
simulating at various pressures (1.0 atm, 50, 100 , 500, 1000, 5000), I
found the pressure vs density relation to be linear, and a regression
line indicated that in order to obtain 0.997 density, one needs to run
at approximately 356 atm. Has anyone else seen this type of behavior? I
have been checking the literature values and all papers I have found
report the opposite trend (i.e., TIP3P appears to be slightly too dense
at 1 atm 298K). I have thought that maybe the box size is too small --
however, a few of the oft-cited original TIP3P papers (1983, 1985) by
the Jorgensen group has used the same system size. I would appreciate it
if anyone has any insights into this. Thank you very much.

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:45:09 CST