[gmx-users] GROMOS vs. GROMACS

[Berk Hess]

When simulating water using a cut-off for the Coulomb
interactions or using a reaction field, heating occurs due
to cut-off effects. This heating is always present and
depends on the cut-off length and the time step.
Using the usual cut-off's (0.9 to 1.4 nm), time step (2 fs)
and berendsen thermostat coupling time (0.1 ps), this
results in a temperature which is 2 to 5 degrees above the
reference temperature, which is visible in Gromacs.

Using the Gromos package this effect is almost invisible.
This is because the Gromos temperature calculation and couping
code use two different temperature definitions, which is not
consistent. Most of the code uses the temperature at the
full timestep (like in Gromacs), but the temperature used
for the scaling is calculated in between two timesteps, where
one has the velocity in a leap-frog algorithm. This temperature
is one average higher than the temperature at the whole timestep,
which causes a larger temperature (down)scaling. This result is
that the temperature in Gromos will be lower. For water simulation
with cut-off it is accidentally very close to the reference
temperature. But the temperature could under certain conditions
even end up below the reference temperature!

The only real solution for heating in water simulations with
cut-off is not to use a cut-off (use PME).
But a temperature difference of a few degrees is much smaller
than the errors in the forcefield.

Berk.



_________________________________________________________________
Send and receive Hotmail on your mobile device: http://mobile.msn.com