What are the differences between EMD and NEMD? How one decides that which one should be used for a run?
I went through some materials available on web but didn't develop a clear concept about these two.
Please help me understanding.
If possible please include some information on available packages where NEMD can be performed.
Thanks.
In EMD you first thermodynamically equilibrate the system. A first sign that the investigated system is in the thermodynamic equilibrium is the convergence of some parameters (E, P, T ..), depending on the simulation ensemble. In NEMD the system is generally evolving after a perturbation. A good discussion about the differences between EMD and NEMD is given in the article "Nonequilibrium molecular dynamics", Wm.G.Hoover, C.G.Hoover, Condensed Matter Physics, 2005, Vol. 8, No. 2(42), pp. 247–260.
In EMD you first thermodynamically equilibrate the system. A first sign that the investigated system is in the thermodynamic equilibrium is the convergence of some parameters (E, P, T ..), depending on the simulation ensemble. In NEMD the system is generally evolving after a perturbation. A good discussion about the differences between EMD and NEMD is given in the article "Nonequilibrium molecular dynamics", Wm.G.Hoover, C.G.Hoover, Condensed Matter Physics, 2005, Vol. 8, No. 2(42), pp. 247–260.
Dear Sayantan,
the majority of the community assumes that equilibrium in an atomic system is defined when the velocity distribution of the system resembles a Maxwellian and the system satisfies the equipartition of energy (1.5kT). MD simulations with stable integration algorithms satisfy this quite easily, which is why that is the preferred definition for most people.
However, the more accurate way of determining whether an MD simulation is in equilibrium is to calculate the phonon modes of the system and to ensure that atomic vibrations follow those phonon modes. If the time required to reach equipartition of energy is shorter than the lifetime of the phonon modes (the time that it takes for these modes to begin interacting with other modes), as might be the case in silicon, then one cannot formally talk about equilibrium. But this procedure is much more cumbersome and seldom used.
Technically speaking, non equlibrium MD is anytime you don't strictly satisfy the equilibrium conditions described above. However, to me, NEMD occurs whenever you have a clear driving force in the system. In most atomistic simulations, driving forces take the form of spatial gradients of simulation quantities such as temperature, concentration, stress, etc. It is relatively easy to impose boundary conditions that can produce these gradients and so it should not be a problem in general to run all kinds of NEMD simulations
- Badges
- Science topic