Dear Axel,
I would like to ask a question about the interactions taken into account to determine the ECI by maps.
In the case where there are 2 sublattices in the file lat.in, the first occupied by atom A and the second by atoms B and C, are the interactions :
between atoms from the first sublattice (A-A)
between atoms sharing different sublattices (A-B and A-C)
included in the ECI ?
If not, are there possibilities to include them ? More generally how is it possible to have further information on the construction of the ECI ?
Thank you very much for your explanations and advise.
Best regards
Interaction between atoms from the first sublattice (A-A) would be accounted for as part of the "empty" cluster.
Interaction between atoms sharing different sublattices (A-B and A-C) would be accounted for as part of the point cluster on the (B,C) site.
In the cluster expansion formalism, there is to separate out A-A or A-B,A-C interactions: there is imply not enough variation in the input configurations to distinguish, e.g., the self-interaction on the (B,C) site and the A-(B,C) interactions.
thank you very much for your answer, I understand better.
Another question occurred to me : I am not quite sure to understand the nature of the ECI of the point cluster determined by maps.
-Is it a constant on the whole concentration range ?
-Is it injected in the Monte-Carlo stage and what is the relation between this quantity and the potential mu ?
Point ecis are not constant, they depend linearly on composition.
They enter the MC energy in the same way as the chemical potentials (they give an additive shift to the chemical potentials).
Thank you again for your answer !
I thought about and it is already much clearer. May I ask you a confirmation ?
What I call "point ECI" is the ECI that is given in the file "eci.out".
As I understood, the energetic contribution of the point cluster depends thus linearly on the concentration because it is obtained by multiplying :
-this point ECI (constant, from "eci.out")
-by the point correlation function, which is a linear function of the concentration.
Is it right ?
I am sorry, I mistyped: the point correlations depend linearly on composition.
The energy expression has a term that is (point correlation)*(point eci). The point eci is constant (and given in eci.out). The point correlation varies.
The (point eci + chemical potential) are multiplied by the point correlation in the MC code.
I hope this is clearer.