ABINIT Discussion Forums

Dear Members of the Forum,

As documented in the description of 'ecut',
http://www.abinit.org/documentation/helpfiles/for-v7.6/input_variables/varbas.html#ecut
or in
[page 2]
the number of planewaves can be different at different kpoints.

I wonder if it is possible to enforce ABINIT to output the non-scf WFK with the same number of planewaves at all k-points.
(perhaps using dilatmx ?) I know there is 'npwwfn', but it only applies for screeening or sigma calculations.

My interest in this is because I wish to implement a Model Hamiltonian starting from the non-scf WFK. At some point I need to perform projections between states at a fixed kpoint k and a set of states at different kpoints (q1, q2, ..). This leds me to read the "cg" that contains the wavefunction coefficients. But then I face the issue that the number of G's is different for different k-points, i.e. the length of cg_nk(G) can be different that the length of cg_mq(G).

If this can not be enforced for nonSCF WFK's, would the following be a suitable solution?
1) to sort out all the cg's, so that the i-th planewave always corresponds to the same combination of reduced coordinates, say [n, l, m], for all cg's.
2) then apply a filter where I remove the highest planewaves,
(those closer to the surface of the sphere (k+G)^2 ),
so that all cg_nk (G) have the same number of G's for any nk

By the way, how does abinit checks orthogonality?
The normalization check does not face this lenght issue, as it requires summing up states at the same k-point, i.e.
sum_spin sum_G | c_k(G,spin) | = 1, for a given k
which is the last eqn of Section 1 of

With regards,
Temok