How to view db.kindx or how to know the k sampling?

[Guangfu Luo]

Dear friends,

(1) After running a BS calculation, I tried to decompose the excitons to KS e-h pairs. But when I analyze the o.exc_weights_at_* file, I do not know the exact value of the k-point under the following line refers to. (Based on the next question, I am not sure if Yambo uses the exact k-points of the KSS file; Meanwhile, when I plot the electron density of KS band by using "ypp -s w" and compare it with that from Abinit, I find Yambo roate the axis by a certain angle).

"Band_V------Band_C------K ibz------Symm.------Weight------Energy" (empty spaces are replaced by dashes)

So I hope to read the SAVE/db.kindx (I did not employ netCDF before). Could you enlighten me how to fullfill this? Maybe we can do some simple calculation which can export all the k-points? Or someone knows the file format and has a tool to read it?

(2) By the way, in the same o.exc_weights_at_* file, I found some k-points under the following line are not included in the KSS file, which the BS calculation is based on. Actually these kpoints are two times denser in each axis than that in the KSS file. Does Yambo do the BS calculation this way, or I misunderstood something?

"#-----------K-point [iku]-----------Weight" (empty spaces are replaced by dashes)


Thanks very much for your help.

[Daniele Varsano]

Dear Guangfu Luo,
in the o.exc_weights_at_* file you find the transition that contributes to the excitation
for more than 5 or 10%. So it is possible that not all the k-point are present there, but
does not means that they are not included in the calculation.

The db.kindx cannot be read itself, but the command yambo -D shoud give you the
headers of all the databases. If you switch to netcdf there are command to read inside
the database files.

Anyway all the useful information of the databases are found in the report file!!
If you have problems, please post your outputs and reports file, we will have
a look into them.

Cheers,

Daniele

[Guangfu Luo]

Dear Daniele,

Thanks very much for you swift reply. Now I can get the k-points that Yambo uses from the report file. The results confirm my former expectation: the k-points in the KSS file (by checking the output file of Abinit) and those in Yambo (by checking the report file of Yambo) are different. I plotted them in the whole BZ and find that they are not equivalent.

The different k-points in Abinit and Yambo are attached for your inspection. The system I tested is graphene and the test-purpose input files for Abinit and Yambo are also attached.


Could you please give me further help?

Sincerely,
Guangfu Luo

[Daniele Varsano]

Dear Guangfu Luo,

in your abinit output file I see this line:

istwfk 1 0 0 0 0 1 0 0 0 0
0 0 0 0

this can cause problems, in the nscf calculations these should be all 1.
Next, even if not at all related with your k-point sampling,
careful with the energy cutoff, you have something like 3.67Ha. I think
you are strongly out of convergences for carbon atoms!!!

Cheers,
Daniele

[Guangfu Luo]

Dear Daniele,

In the Abinit input file, I specify the istwfk to be 1 for all kpoints. I check the KSS file by using the cut3d tool in Abinit, and all the istwks are 1 indeed. More importantly, cut3d shows that the header of the KSS file contains the same k-points as those in the Abinit output file.

So the problem should be on the Yambo side. Yambo does not employ the k-point sampling in the KSS file---at least in this case.

For the low ecut, I just decreased it to a low level, so you can repeat all the results in minutes. Anyway, thanks for your kindness.