Thermal conductivity of system with temperature gradient

Regarding the accuracy of the results is there something I should take into account? For example when using Green-Kubo relations you have to allow significant time until the correlation function decays to zero and the thermal conductivity converges to a value.

With the NEMD method, it sounds that I can basically calculate the thermal conductivity as soon as the system reaches equilibrium. Is there anything I should be aware of?

With NEMD your system never reaches equilibrium. You keep it away from equilibrium using the thermostats. It will reach a non-equilibrium steady state, in which you get a constant temperature profile. Once you've got a steady gradient and heat flux, you can work out the conductivity.

Perfect.

Thanks you very much for your quick and helpful answers

Mike