Problem in GW to BSE

[Flex]

Hello,

I am working on a monolayer MoS2 system.
I have done several GW and BSE calculations successfully (from pwscf results), and I'm attempting for the first time to use my GW results into a BSE calculation. For that, I used the same parameters as a previously successful BSE from DFT calculation, added

KfnQPdb= "E < SAVE/ndb.QP"

And removed the "em1s" tag because it is supposed be already done by GW. I verified my GW results and they seem to fit.

So when I try to do this, I get

- "<05s> [WARNING] The system is a metal but Drude term not included. " Strange, because the GW bands fit well to the semi-conductor MoS2 and are not metallic

Then

- "P001: [ERROR] STOP signal received while in :[05.01] Main loop
P001: [ERROR] Error in parallel wf distribution"

And the calculation fails

Is this some I/O error ? Bad parameters in BSE ? Are my GW results not so good ?

I'm attaching the ins and outs of these, and the bands I get from GW.

[Flex]

Only 3 attachment per post, so here are the bands

[Daniele Varsano]

Dear Flex,
can you please post the report file of your calculations?
Next, the band structure you posted is the KS structure. You can interpolate the QP structure in order to see if something is wrong with your QP correction by adding

Code: Select all

 GfnQPdb= "E < SAVE/ndb.QP

in the ypp input files, but you can also have a look to the QP output/report in order to see if something odd happened. The Fermi level is recalculated after the correction is applied, and the system is recognized as a metal after the correction.

Best,
Daniele

[Flex]

I did the ypp with the QP (QP.png, compared to KP.png which is without) interpolation and something seems to be wrong. It still has a gap (on the path I chose) but seems very different from the original bands, that look more like KP.png. (with a direct gap at the corners of the hexagonal BZ)

I also looked at the GW report file (in the next post, split in three because of the filesize limit). It seems the system actually is metallic after the calculation but I have no idea why.
I'm still quite new to this. Should I change some parameters ? Is the precision or the convergence not good enough ?

[Flex]

The report file

[Daniele Varsano]

Dear Flex,
the Qp correction seems to me not be the responsible to turn the system metallic.
The problem is most probably when reading the QP energies and plugging them in the BSE kernel.
From your previous input file I can see you are using a rather old version of yambo (3.4.1). I strongly suggest you to update to a newer version (3.4.2 or 4.0.2) as we fixed several issues since then and start a new calculation from scratch.
Next, are you sure that you will need all these bands for the BSE? May be you can consider to not include the fist deep bands in the BSE calculations, in this way you will save several computational time avoiding to calculate the QP correction for them.
If the problem persist, please post here again, and we will have a deep look to it, in order to spot the problem for us it is always better to consider the newer release.

Best,
Daniele