GW+BSE Calculations

Problems running VASP: crashes, internal errors, "wrong" results.

Moderators: Global Moderator, Moderator

Post Reply
Message
Author
dscullion12
Newbie
Newbie
Posts: 10
Joined: Wed Nov 25, 2015 2:38 pm
License Nr.: 5-2233

GW+BSE Calculations

#1 Post by dscullion12 » Mon May 15, 2017 10:09 am

Dear VASP Users,

I have been carrying out GW+BSE calculations on a semi-conducting material. When carrying out the BSE calculation I use the WAVECAR and WAVEDER files from the groundstate calculation and the WFULLxxxx.tmp files from the GW calculation. However, the GW calculations has increased the energy of the top of the valence band such that it is now larger than the fermi level of the groundstate calculation.

When carrying out the BSE calculation the material is being treated, wrongly, as a metal.

Would anyone have a solution to this problem?

Kind regards,

Declan

gbal
Newbie
Newbie
Posts: 17
Joined: Mon Oct 12, 2009 2:25 pm
License Nr.: 5-82
Location: Stockholm, Sweden

Re: GW+BSE Calculations

#2 Post by gbal » Thu May 18, 2017 10:22 am

Hi there,
First i would say that this would fit better for "physics questions" than "bug reports", but here we are.

Any special reason to use LDA/GGA WAVECAR and WAVEDER?

I normally calculate BSE using the outputs from the GW calculation.

Maybe you wanna make TDDFT (just don't have the GW files there) if you need to keep using the ground-state files.

dscullion12
Newbie
Newbie
Posts: 10
Joined: Wed Nov 25, 2015 2:38 pm
License Nr.: 5-2233

Re: GW+BSE Calculations

#3 Post by dscullion12 » Fri May 19, 2017 12:10 pm

Thank you for the quick reply.

I can see reasons for using the GW WAVECAR files instead of the groundstate when solving the BSE. Thank you for clearing this up.

I would have one more question though. Depending on the parallelisation method I use (KPAR or a combination of NPAR and NCORE) I get different results from my G0W0 calculations. For example, using KPAR = 2, NPAR = default, NCORE = default I obtain a bandgap of 3.19 eV. If, however, I used KPAR = default and NPAR = 8, NPAR = 96 I obtain a bandgap of 5.23 eV.

Nothing has been changed between the two calculations apart from the method of parallelisation.

Any help would be greatly appreciated.

Kind Regards,

Declan

gbal
Newbie
Newbie
Posts: 17
Joined: Mon Oct 12, 2009 2:25 pm
License Nr.: 5-82
Location: Stockholm, Sweden

Re: GW+BSE Calculations

#4 Post by gbal » Mon May 29, 2017 10:16 am

dscullion12 wrote: Nothing has been changed between the two calculations apart from the method of parallelisation.
Are you sure about this?

When you change parallelization tags, you change several things, including number of bands and readability of files depending on it.
My first bet would be that you are with different settings. And even in the case that you are reading the files properly, this is already a different setup that will influence the results. GW parallelization works in a different manner than standard DFT or HSE, prepare the calculation accordingly.

call_me_Al
Newbie
Newbie
Posts: 20
Joined: Tue Oct 04, 2016 7:05 pm
License Nr.: 5-2273

Re: GW+BSE Calculations

#5 Post by call_me_Al » Sun Nov 19, 2017 4:17 pm

gbal wrote:Hi there,
First i would say that this would fit better for "physics questions" than "bug reports", but here we are.

Any special reason to use LDA/GGA WAVECAR and WAVEDER?

I normally calculate BSE using the outputs from the GW calculation.

Maybe you wanna make TDDFT (just don't have the GW files there) if you need to keep using the ground-state files.

hi gbal,

I was wondering if the TDDFT is currently supported in VASP? I know there is some TDHF algorithms available in VASP with limited support, but I am eagerly awaiting a TDDFT extension for solid state optical calculations. Maybe you have some more information on this.

thanks

gbal
Newbie
Newbie
Posts: 17
Joined: Mon Oct 12, 2009 2:25 pm
License Nr.: 5-82
Location: Stockholm, Sweden

Re: GW+BSE Calculations

#6 Post by gbal » Thu Jan 18, 2018 3:16 pm

Hi there,
About the TDHF and TDDFT, i have the same question as you!

Post Reply