simulating dynamic properties

Hi Lammpers:

I am trying to simulate what I think of as dynamic properties. Here is my example: I have a periodic boundary condition box full of polymer. Chains of 10 monomers each. I am dropping the temperature by 0.1 instantaneously, from 0.7 to 0.6 say. I want to watch the energy dip as the systems relaxes into the lower temperature. I expect some sort of exponential decay in energy. I cannot seem to get away from the damping qualities of the integrator (I use npt here). The damping drives the system. Whatever damp I choose: 10, 100, 200 in each case the energy is driven by the wavelength chosen by the damp parameter.

The problem is the barostat and thermostat drive the dynamics, and the only statistics I can rely on are the ones derived from properties associated from the equilibrium.

One question is, how do I get away from the stat damping?

More deeply, I seem to be at a point here where questions of dynamics (pre eqilibrium) just cannot be answered by molecular dynamics?

Thanks for your thoughts.

You can try fix AtC, the thermal coupling lets you approximate non-equilibrium boundary conditions. It is essentially a complicated thermostat since it applies damping as well.

How about just rescale the temperature (instantaneously), then

run NVE (no thermostat) and see how things relax.