[gpaw-users] Questions regarding dielectric function in GPAW

[Thomas Olsen]

Hi

Let me just address your last question. The GW method is good for calculating single-particle excitations - the band structure. For the dielectric function
you need a good description of the two-particle excitations, which is typically accomplished with the Bethe-Salpeter equation (BSE). So a state of the art calculation of \epsilon will involve doing a BSE calculation on top of GW bands. However, the calculations are quite heavy so it is usually only possible to obtain the imaginary part of \epsilon in a certain frequency range using this method. However, for conductors I think there will not be a lot of difference between BSE and RPA. You could perhaps include the GW energies in the response calculation if you like - you would then just have to set the eigenvalues in
the calculator object before doing the response calculation

Regarding the setups I do not think it will make a lot of difference if you for example use PBE setups for a PBEsol calculation (just copy the PBE setup into a PBEsol setup). I am not sure though. Alternatively you can generate them easily with the default settings.

I think there is a beta version of the response code that includes SOC, but I am not sure if it has made it to the developer version yet. Morten Gjerding?

/Thomas



________________________________________
Fra: gpaw-users-bounces at listserv.fysik.dtu.dk <gpaw-users-bounces at listserv.fysik.dtu.dk> på vegne af Matej Bubaš via gpaw-users <gpaw-users at listserv.fysik.dtu.dk>
Sendt: 28. august 2018 16:49
Til: gpaw-users at listserv.fysik.dtu.dk
Emne: [gpaw-users] Questions regarding dielectric function in GPAW

Dear GPAW users,

I am Matej Bubaš, currently starting my PhD at Ruđer Bošović institute.

I am researching the optical properties of solid conductors, and have
recently started to use GPAW for band structure and dielectric
function calculations.
So far I am pleasantly surprised with the possibilities and the
usability of GPAW.

Recently I've had a couple of dilemmas which I was hoping you could
help me with.

I was trying out different functionals, since they significantly
affect the band structure and the dielectric function. As far as I
understand, in GPAW, the only available PAW setups at the moment are
for LDA, PBE, revPBE, RPBE and GLLBSC functionals. I suspect using
PBE0 and especially PBEsol could prove useful, however there seem to
be no PAW setups available for them. From the e-mail correspondence in
the following link it seems to me that generating setups is
disincentivized:
https://listserv.fysik.dtu.dk/pipermail/gpaw-users/2014-August/002870.html I
was thinking of trying another software, but I would like to keep
using GPAW, so I would appreciate if you could tell me if there are
already available setups for PBEsol and possibly other functionals. If
not, do you think that trying to create my own setup could prove to
be  very problematic, considering I am currently inexperienced in
computational physics?

Another way to improve the dielectric function could be by
implementing spin-orbit coupling in the calculation. For the band
structure calculation the spin-orbit coupling correction is easily
implemented in GPAW, but as far as I can see the correction can not
be, later on, used for the dielectric function calculation. Do you
know if there is currently a way to implement spin-orbit coupling in
the dielectric function calculation using GPAW?

Finally, I would appreciate your opinion on whether it could be
possible to use GW approximation to improve the dielectric function in
GPAW. In the tutorial I've seen it used to improve the band
structure.  However it is not clear to me whether the results obtained
using GW  calculator object could be used in a simple way (with a
simple script,  without changing the source code or writing a lot of
new code for  that) to obtain the dielectric function. Looking at the
source code I  noticed some shared elements, for example, both
calculations using the  chi0 object (class for calculating the density
response function), but  I don't see a straightforward way of
connecting them.

Many thanks and best regards,

Matej Bubaš, mag. chem.

_______________________________________________
gpaw-users mailing list
gpaw-users at listserv.fysik.dtu.dk
https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users