ABINIT Discussion Forums

Dear ABINIT Developers and Users,

ABINIT offers Hartree-Fock calculations withing the GW framework by setting gwcalctyp = 5, 15, or 25
http://www.abinit.org/documentation/helpfiles/for-v6.12/input_variables/vargw.html#gwcalctyp

And a recipe for such a calculation is given almost at the end of
http://www.abinit.org/documentation/helpfiles/for-v6.12/tutorial/lesson_gw1.html#10

which is exemplified in the test v6#89, provided by F. Bruneval. See Section D.8 of
http://www.abinit.org/documentation/helpfiles/for-v6.2/release_notes/release_notes.html/?searchterm=Hartree-Fock

I have a couple of questions about the way this is implemented --coded-- in ABINIT.

First, in considering the approximation to the self-energy given by Sigma=GW,
if we replace the screened Coulomb interaction W by the bare Coulomb interaction,
and if then we put the resulting Sigma in the Dyson equation,
the latter gets reduced to the Hartree Fock equation.
This is described in "Conceptual Foundations of Materials: A standard Model
for Ground and Excited State Properties", Chapter 2 by Steve G. Louie,
page 28. Elsevier 2006.

Is this actually the way ABINIT implements Hartree-Fock calculations
( replacement of W by bare Coulomb ) ?

Second, in a true Hartree-Fock calculation, the exchange interaction is computed exactly,
whereas correlation effects beyond exchange are exactly ignored.
Would this be fulfilled for any material in ABINIT?, either for pseudos or PAW?

The latter question arises because of the following:
suppose I perform a Hartree-Fock calculation in ABINIT (gwcalctyp = 5, 15, or 25)
either using a pseudopotential file (like TM) or a PAW dataset.
Such files include an approximation of exchange-and-correlation
effects in the exchange-correlation functional.
How ABINIT ensures that no correlation is included in a Hartree-Fock calculation?

My purpose is to perform true HF calculations, avoiding all
correlation effects beyond exchange, using an all-electron PAW file I have.

Best regards,
Temok