Ben Hourahine
@thomas-niehaus this is possibly working for Arpack, and for the test case it is around 7x faster to get the 20 states near to 6eV compared to solving for ~70...
This is a useful feature from the user side, for example if the exported viewer does not understand periodicity (or a code feature requires the atoms in the central cell)....
@awvwgk still working on this, a few more days to put it back together. To check, you want the Born charges (i.e. the mixed 2nd derivative of energy wrt to...
The derivative of forces wrt to applied electric field is equivalent, and more efficiently calculated for systems containing > 3 atoms. Both give the Born charges.
@jubich Is this ready to review? (as still listed as draft). Also please resolve the conflicts against main.
Oh, and regression testing for the original script.
Probably needs error propagation upwards through the C API to fix (internally at the moment it hits call error() in Fortran but this doesn't cleanly shutdown over the ctypes interface).
The two argument form does, but yes probably more of an enhancement.
Yes, its all legal Fortran. But as a pattern its more of a learning curve and cognitive overhead for new developers/less Fortran experienced developers (who need to know to recognise...
Partly addressed by #491