Dear Axel,
could you give me feedback if I got the essentials of the theory right?
From what I understand the quasi-harmonic approximation is implemented in ATAT for the calculation of phonon energy contribution (constant pressure heat capacity) to the total free energy. So the cluster expansion can be extended onto the energetics at T>0K.
Moreover I understand that the code uses the finite differences method for the calculation of the force constants, right? So for each structure one needs to run calculations on perturbed structures, which ATAT will generate when issuing the command fitfc and which is actually the exhaustive part of the cluster expansion.
Siau