Just as an additional comment, we currently don't support calculating thermal conductivity via the Green-Kubo approach, because we don't have the atom resolved potential energy implemented. That might be something to be included in the next release. As an alternative to NEGF, you can use Non-equilibrium MD simulations, often called the "direct method", see
http://docs.quantumwise.com/tutorials/interfacial_thermal_conductance/interfacial_thermal_conductance.html.
The constant volume heat capacity can be calculated from MD, assuming that the system behaves purely harmonic, by calculating the vibrational density of states from the VelocityAutocorrelation MD analysis object (
http://docs.quantumwise.com/manuals/Types/VelocityAutocorrelation/VelocityAutocorrelation.html#velocityautocorrelation-c), and then integrating over the VDOS to get the heat capacity as shown in the attached script.
To get reliable results, it is important that you carefully equilibrate the system (using NVT Nose Hoover) and that you run the actual MD simulation in NVE with a small log interval of around 10 and I would recommend a smaller time step than normal, e.g. 0.5 fs. You typically don't need to run the simulation very long, 40 000 steps is normally enough. Check in the MD-Analyzer that the vibrational density of states looks reasonable (use Velocity Autocorrelation and the "Show vibrational DOS" checkbox).
For a non-harmonic system, you can use the approach in the paper, mentioned by Petr, by running an MD simulation and very slowly increasing the temperature. The slope of the energy-temperature curve will give you the heat capacity.
