GW spectrum

[WangFuyou]

Dear yambo developers,
I'm using yambo to get a spectrum of GW. But I got a lot of o.ep_q_ip files. I know that the second column of data represents the imaginary part of the dielectric function. But I don't know which file's data is what I need.
I uploaded my input file and my output file. I hope I can get your help.

[Daniele Varsano]

Dear WangFuyou,

in your input you set:

Code: Select all

% QpntsRXd
  1 | 30 |                           # [Xd] Transferred momenta
%

so you are asking for the spectrum for several transferred momentum. If you are interested in "optical" spectrum, you need to look at the long wavelength limit q~0, while other q point spectrum are useful in the interpretation of electron-energy loss experiments where a finite momentum can be transferred.

In Yambo this limit is always the q index number 1, so you need to look at the o.eps_q1_ip file.

In general if you are interested in optics only you can set in the input:

Code: Select all

% QpntsRXd
  1 | 1 |                           # [Xd] Transferred momenta
%

so avoiding calculating the spectrum for other momentum.

Best,
Daniele

[WangFuyou]

Dear Daniele,
Thanks for your reply, now I understand the meaning of this parameter. But when I used the imaginary part of the dielectric function to plot the spectrum, I found that it was almost the same as the result obtained by DFT. I'm pretty sure there's something wrong with that. But I did set KfnQPdb = E <./FULL/ndb.QP.
I want to know what caused it. I guess it's a convergence problem. But I don't know how to tell that the convergence requirement is met, because I don't know how to get the band gap of GW.
I look forward to your guidance. Thank you very much.

Yours,
WangFuyou

[Daniele Varsano]

Dear WangFuyou,

it seems the QP database has not been read. If you inspect the report file, at the end you find a mirror of the input variable used by Yambo in the run, and you can recognize that the KfnQPdb was not read.

Here there are two issues:
1) the database path should be inserted with quotes: "E < ./FULL/ndb.QP"
2) You are not in the BSE runlevel, and for a linear response calculation the right variable is XfnQPdb and not KfnQPdb

The general suggestion is to generate input by command line and not by hand:
in your case, the command line is:

> yambo -o c -V qp

The -V option adds verbosity in your input file, showing all the optional variables.

>yambo -h
gives you a help on how to generate input files.

Best,
Daniele

[WangFuyou]

Dear Daniele,
Thank you for your answer. I ran the calculations again as you suggested. But it didn't work.
I found such a warning in the r_optics_dipoles_chi_01 file :
[WARNING]I/O problems in reading QP database ./FULL/ndb.QP
I wonder if the results of the QP were not successfully read.Looking forward for your help.

yours,
WangFuyou

[Daniele Varsano]

Dear WangFuyou,

indeed, the QP file was not read. Do you have any error message in the log files l_* in the LOG directory?
This can help in spotting the problem.
Best,
Daniele