Calculating Green-Kubo based thermal conductivity using ACE potential

[rfhari]

Hi, I am interested in using ACE to develop an interatomic potential for AlGaN. I plan to use this potential to calculate Green-Kubo based thermal conductivity from EMD simulations, as show in Eq. 1 of this paper. I am currently using the Pacemaker package for training the potential and LAMMPS for calculating heat flux. It appears that the energy_corrected (i.e. the reference or cohesive energy) is used for the fitting process.

So, I’m wondering if the ACE-predicted energies are adjusted to the actual energies (by adding the free energies of individual atoms) during MD simulations. If not, could this result in an underestimation of the potential energy flux (a component of the total heat flux), even though, I believe, the force flux and kinetic energy flux (other components of heat flux) would remain unaffected?

[cortner]

Sorry nobody answered you. It seemed a quite application specific question which I won't be able to help with. I also just noted that you are using a different package. So why ask here?

[rfhari]

Thanks for the reply! Yeah it makes sense I'll close the issue here.