[nedtaylor]

Use of formation energy as the weighting to sum distribution functions (DFs) to generate the generalised distribution function (gDF) is flawed due to the incorrect comparison of structures with differing stoichiometries.

This is extremely prevalent when considering choice of reference energies. If using isolated atom energies for a binary compound material, then one element species will have a much more favourable formation energy for its bulk (i.e. carbon) than the other (i.e. oxygen). This means that the second element in its bulk phase will contribute nothing to the gDF, which is not what should happen.

[nedtaylor]

Another addition made to this branch is that the formation energy of the host structure is subtracted from the generated structures. This may be counterintuitive, but it is to show the favourability of placing an atom in the specific host structure, without any bias of unfavourability of the initial host (i.e. energy cost of termination, etc).

[nedtaylor]

Another addition made to this branch is that the formation energy of the host structure is subtracted from the generated structures. This may be counterintuitive, but it is to show the favourability of placing an atom in the specific host structure, without any bias of unfavourability of the initial host (i.e. energy cost of termination, etc).

Actually, it negates the energy of formation of the host structure. This is the important value to subtract as that contains the formation energy of the terminations.

[shepplestone]

In principle approved, deep code review due 12/10/2024