output file of BSE

[sdwang]

Dear all"
I have calculate my 2D rectangle unitcell including 4C atoms and 2O atoms with pwscf v4.1.2, then I do a BSE calculation.I type >yambo -o b -b -y h,and edit input file as:
optics # [R OPT] Optics
bse # [R BSK] Bethe Salpeter Equation.
em1s # [R Xs] Static Inverse Dielectric Matrix
bss # [R BSS] Bethe Salpeter Equation solver
BSresKmod= "xc" # [BSK] Resonant Kernel mode. (`x`;`c`;`d`)
% BSEBands
1 | 18 | # [BSK] Bands range
%
BSENGBlk= 1 RL # [BSK] Screened interaction block size
BSENGexx= 2027 RL # [BSK] Exchange components
% QpntsRXs
1 | 14 | # [Xs] Transferred momenta
%
% BndsRnXs
1 | 18 | # [Xs] Polarization function bands
%
NGsBlkXs= 1 RL # [Xs] Response block size
% LongDrXs
1.000000 | 0.000000 | 0.000000 | # [Xs] [cc] Electric Field
%
BSSmod= "h" # [BSS] Solvers `h/d/i/t`
% BEnRange
0.00000 | 10.00000 | eV # [BSS] Energy range
%
% BDmRange
0.10000 | 0.10000 | eV # [BSS] Damping range
%
BEnSteps= 100 # [BSS] Energy steps
% BLongDir
1.000000 | 0.000000 | 0.000000 | # [BSS] [cc] Electric Field
%
Then type >yambo -N, it seems to go well .and I got the optical spectrum of it, but this result is the same like the RPA one.The two OAS agree very well. Is the result believable?
The RPA input file is:
optics # [R OPT] Optics
chi # [R CHI] Dyson equation for Chi.
% QpntsRXd
1 | 14 | # [Xd] Transferred momenta
%
% BndsRnXd
1 | 18 | # [Xd] Polarization function bands
%
NGsBlkXd= 1 RL # [Xd] Response block size
% EnRngeXd
0.00000 | 10.00000 | eV # [Xd] Energy range
%
% DmRngeXd
0.10000 | 0.10000 | eV # [Xd] Damping range
%
ETStpsXd= 100 # [Xd] Total Energy steps
% LongDrXd
1.000000 | 0.000000 | 0.000000 | # [Xd] [cc] Electric Field
%

What is the problem??
When I calculate bulk Si, the speed is slow.But my unitcell including 6 atoms is very quick as RPA procedure. I do not know why??

Thanks in advance !!
SDWang

[Daniele Varsano]

Dear Shudong,
I don't know what are you calculating, but note that:

In BSE you are considering just one block in the screening interaction.

BSENGBlk= 1 RL # [BSK] Screened interaction block size

In the RPA you are not including local fields:

NGsBlkXd= 1 RL # [Xd] Response block size

moreover:
when a BSE calculation is performed, you should consider quasiparticle
energies, so in practice you have to correct your eigenvalues by a previous
GW calculation or some empirical corrections.

I suggest you to take a look to the documentation.

Best,

Daniele

[andrea marini]

I suggest you to take a look to the documentation.

... and I strongly invite you to run and study (critically) ALL the tutorials before running any production run.

[sdwang]

Thank you for your reply.

[sdwang]

Dear developers:
After complishing my BSE calculation, I got some data as following :
E/ev e/Im e/Re eo/Im eo/Re e`/Im e`/Re

I wanna know the difference among e/Im,eo/Im and e`/Im, Which is the last data I used to plot absorption spectrum?

Thanks!
SDWang

[myrta gruning]

Hallo Shudong Wang,

e/Im gives you the absorption spectrum within the BSE level of theory.
eo/Im is the absorption spectrum from the independent particle approximation
The comparison of these two curves gives the measure in which many-body effects are important.

e`/Im appears when you are solving the BSE Hamiltonian with Haydock (sec 3.2 of the yambo paper).
This is an iterative algorithm. At each iteration the algorithm produces a better approximation for the spectrum. The algorithm stops when the diff between the spectrum at two successive iteration is lower than a threshold.
So e/Im is the spectrum for the last iteration, and e`/Im the one for the second-last iteration. In principle they should be (almost) indistinguishable, if the algorithm is well converged.

Regards,
m

[sdwang]

Dear Shudong,
I don't know what are you calculating, but note that:

In BSE you are considering just one block in the screening interaction.

BSENGBlk= 1 RL # [BSK] Screened interaction block size

In the RPA you are not including local fields:

NGsBlkXd= 1 RL # [Xd] Response block size

moreover:
when a BSE calculation is performed, you should consider quasiparticle
energies, so in practice you have to correct your eigenvalues by a previous
GW calculation or some empirical corrections.

I suggest you to take a look to the documentation.

Best,

Daniele

Thank you!
I have another quetion about scissor in GW calculation.
According to your opinions, before I have a BSE calculation,I have to firstly perform a GW process(Is it right?), and I would get the quasiparticle
energies. Is the "scissor" in BSE=E_QP-E_LDA???? But in GW calculation, it also have the option "scissor", how can I determine this when I perform GW calculation??
Thanks!

[Daniele Varsano]

Is the "scissor" in BSE=E_QP-E_LDA???? But in GW calculation, it also have the option "scissor", how can I determine this when I perform GW calculation??

Right, this is the definition of the scissor. About your second question, just ignore it, the default value is
0| 1| |1

which means, no scissor.

Best,

Daniele

[sdwang]

Thank you very much

[haseebphysics1]

Hi, dear Yambo team,

I'm new in using Yambo code. My first objective currently is to find optical properties of bulk semiconductors solving BSE. I'm having a hard time locating the output file of the calculation of BSE! I followed the Yambo research paper (Yambo: An ab initio tool for excited state calculations) for command-line options. Moreover, I am also attaching the input and output files (I found relevant). The name of the output file which is mentioned in the paper "o.eps_q001-bh" can not be found in the working directory or any other file containing the data to plot, hence I'm uncertain from where to get data to plot the absorption curve! Maybe it can be due to the reason I am not passing the input commands for parallel calculation correctly?

2nd Q.tar.gz

Anticipating your kind response,

Regards,
Haseeb Ahmad,
LUMS, Pakistan.