Other codes

Codes other than FHI-aims, which employ a localized atomic basis set and the RI formalism can use the main routines provided by CC-aims to obtain CC4S input after a HF- or KS-SCF calculation. To do that, the 2 central subroutines of CC-aims CC4S_init_interface and CC4S_interface_driver are necessary.

After having obtained the converged ab-initio quantities from a HF- or KS-SCF calculation, CC4S_init_interface can be called. This routine takes all relevant quantities, which describe the calculated system and the calculation settings, including the number of k-points, the number of basis functions etc... From these quantities CC4S_init_interface calculates how the matrix- and tensor-quantities, from which CC-aims computes the Coulomb vertex, have to be distributed over the MPI tasks. This information is returned by the subroutine in the form of task-local index-bounds, and Scalapack-specific quantities.

The returned information should suffice to (re-)distribute the relevant matrix- and tensor-quantities appropriately for CC-aims.

After the distributed quantities, including HF/KS-eigenvectors, RI-expansion coefficients etc., have been obtained, they can be passed to the CC4S_interface_driver subroutine. This subroutine will perform the calculation of the Coulomb vertex and generate all the files necessary for CC4S.

In summary, the calling algorithm should look like this (pseudo-code):

---

prepare quantities which describe the calculation setting (number of states, number of k-points etc.)
...
call CC4S_init_interface(...)
...
calculate + distribute or just re-distribute matrix-/tensor-quantities (RI-coefficients, KS/HF-eigenvectors etc.)
...
call CC4S_interface_driver(...)
...
finish up/clean up/deallocate

---