Maximum accuracy of computed spectra

[Zeila]

Dear yambo-experts,

I'd like to know what is the maximum resolution of the optical spectra calculations (with yambo - of course).
I.e., assuming that the calculations are well converged, what is the minimum energy separaration of two peaks that can be resolved in the computed spectrum?
In other words, is there any underlying limit due to the theoretical and computational methods? and what are them?

Thanks!
Zeila


---------------------------------------------------------------------------------------------
Dr. Zeila Zanolli

Université Catholique de Louvain (UCL)
Unité Physico-Chimie et de Physique des Matériaux (PCPM)
Place Croix du Sud, 1 (Boltzmann)
B-1348 Louvain-la-Neuve, Belgium
Phone: +32 (0)10 47 3501
Mobile: +32 (0)487 556699
Fax: +32 (0)10 47 3452
e-mail: zeila.zanolli@uclouvain.be
web: http://www.nano2hybrids.net/
---------------------------------------------------------------------------------------------

[andrea marini]

I'd like to know what is the maximum resolution of the optical spectra calculations (with yambo - of course).
I.e., assuming that the calculations are well converged, what is the minimum energy separaration of two peaks that can be resolved in the computed spectrum?
In other words, is there any underlying limit due to the theoretical and computational methods? and what are them?

Ciao Zeila,

as far as I know there is no theoretical limit to the resolution you can get in the optical spectra. Of course, instead, there is a computational that can be estimated by assuming parabolic dispersion in order to link the step of the BZ sampling with a minimal energy resolution. Roughly I'd say that your energy resolution goes like 1/(Number of BZ k-points).

In Yambo there are some tricks that allow you to increase the energy resolution. One is the energy accumulation that allows you to group together the electron-hole transition when using a large number of k-points. This is implemented in the GPL version. Another, instead, uses a Random Integration method to integrate the Green's function while keeping the number of wacefunctions fixed. The latter is NOT yet released in the GPL.

Andrea

[Zeila]

Ciao Andrea,

thanks for the answer.

My question arose since for DFT total energies there is an intrinsic limit of the accuracy of ~10 meV. Then I was wondereing if there was something analogous for optics.

According to your experience and using the best yambo tricks, would it, for instance, be possible to resolve two peaks ~50 meV apart?

Thanks!
Zeila

[wlyim]

as far as I know there is no theoretical limit to the resolution you can get in the optical spectra. Of course, instead, there is a computational that can be estimated by assuming parabolic dispersion in order to link the step of the BZ sampling with a minimal energy resolution. Roughly I'd say that your energy resolution goes like 1/(Number of BZ k-points).

In Yambo there are some tricks that allow you to increase the energy resolution. One is the energy accumulation that allows you to group together the electron-hole transition when using a large number of k-points. This is implemented in the GPL version. Another, instead, uses a Random Integration method to integrate the Green's function while keeping the number of wacefunctions fixed. The latter is NOT yet released in the GPL.

Andrea

I am studying the k-point sampling effect on the smoothness of imaginary part of dielectric function (graphene system). And the smoothness is very sensitive to the choice of k-mesh.

May I ask if there is a trick like tetrahedron method inside Yambo? Is "CGrdSp" also working on my case? I just do "yambo -b -o c"

Many thanks in advance!

Dr. William Yim
Institute of High Performance Computing, Singapore

[claudio]

I am studying the k-point sampling effect on the smoothness of imaginary part of dielectric function (graphene system). And the smoothness is very sensitive to the choice of k-mesh.

May I ask if there is a trick like tetrahedron method inside Yambo? Is "CGrdSp" also working on my case? I just do "yambo -b -o c"

Dear William

at present tetrahedron is not implemented in Yambo, to speed up calculations convergence versus the number of k-points try to have a look here:

http://www.yambo-code.org/doc/inputs/ypp_rim.php

Claudio

[wlyim]

Thanks Claudio!

I will be happy to know what unit of the dielectric function is. Is it in S.I. unit?