I am simulating the supercooled water at 246K and 1bar with using TIP5P model, however, there exists a clear drift after 20ns in my output trajectory file. So where is the problem and whether the trajectory is still valid.
Here shows how i produce the trajectory:
Firstly, I create the 1000 molecules by:
it should be noted that, the tip5p.mol includes:
then i begin producing the equilibrum configuration with:
the total energy and potential energy during this progress has shown equilibrium as following:
while the trajectory after 20ns shows as following(speed up version):
when the trajectory during 20ns shows as following(speed up version):
Hello,
Please follow the rules of the forum. Posting images of input script are a terrible idea, because nobody can test your code.
What you see is a common issue in MD. You can cancel the drift by using the fix momentum. You can also cancel initial momentum using the velocity command.
Best
Simon
I am not sure whether fix momentum will work, since it operates on atom velocities, however the rigid fixes maintain position, orientation and corresponding velocities of the rigid bodies and generate the atom position and velocities from that.
When using the velocity command (recommended), you have to make certain that the rigid fix is properly re-initialized (see its documentation). Removing the center of mass momentum after equilibration is a good practice, since the drift can form more quickly during equilibration.
In general, the drift will eventually appear for almost all bulk simulations, but it can be delayed by reducing the noise from errors. The settings for kspace convergence is insufficient for a variable cell simulation (1.0e-6 is often suggested as a conservative choice), also the timestep may be on the larger side for water with fix rigid (try 0.25 fs) and your cutoff could be larger, too.
