@MSallermann is working on an FFT-based DDI implementation.
Potentially, for more general cases (irregular distributions of atoms), FMM might be implemented as well.
An option to select the DDI implementation needs to be added to input/options for the Heisenberg Hamiltonian.
For generality, the default should be FMM (if/when implemented).
Potentially, an "auto" type would make sense, which chooses the implementation based on boundary conditions and geometry information (e.g. small number of basis cells relative to total number of atoms).
@MSallermann is working on an FFT-based DDI implementation. Potentially, for more general cases (irregular distributions of atoms), FMM might be implemented as well.
An option to select the DDI implementation needs to be added to input/options for the Heisenberg Hamiltonian. For generality, the default should be FMM (if/when implemented). Potentially, an "auto" type would make sense, which chooses the implementation based on boundary conditions and geometry information (e.g. small number of basis cells relative to total number of atoms).