New pseudos and coulombian cut

Can't find a suitable forum for your query? Post it here.

Moderators: Daniele Varsano, andrea marini, Conor Hogan, myrta gruning

Re: New pseudos and coulombian cut

Postby martinspenke » Fri Mar 31, 2017 2:42 pm

Dear Daniele,

Thanks a lot. It is now much clear.

However, I think I have to probably redo calculations because the RIM/Bare deviations
for the q's at the border are larger than 10^-03, as you suggested. But maybe the current deviation can still be considered as okay.

Yes, particle in a box, increasing the vacuum, ie the box in the same energy window, you will find more and more unbound states.
In Yambo you can also define the number of states as energy range, so the number is defined consequently, but me personally do not find it very practical


I see, then I wonder why people investigating convergence on MoS2 https://arxiv.org/pdf/1311.1384.pdf take only 200 bands for vacuums of up to 60 angstrom!
This is somehow irritating.

Bests
Martin
Martin Spenke, PhD Student
Theoretisch-Physikalisches Institut
Universität Hamburg, Germany
martinspenke
 
Posts: 149
Joined: Tue Apr 08, 2014 6:05 am

Re: New pseudos and coulombian cut

Postby Daniele Varsano » Fri Mar 31, 2017 2:48 pm

Martin,
my suggestion 10^-3 was clearly very indicative. You can maybe run a simple check on the exchange self-energy, which is rather fast, and see if you are satisfied with the precision.

ake only 200 bands for vacuums of up to 60 angstrom!
This is somehow irritating.


Literature is full of non-converged calculations :)
Best,

Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/
User avatar
Daniele Varsano
 
Posts: 1985
Joined: Tue Mar 17, 2009 2:23 pm

Re: New pseudos and coulombian cut

Postby Daniele Varsano » Mon Apr 03, 2017 8:14 pm

Dear Thierry,
just a comment on what I wrote you above.
The multi-projector ONCVPSP pseudos are supported only if you perform the ground state calculations using quantum-epsresso, as in the old versions of abinit the multi projectors are not stored in the KSS file. New release of abinit instead are not anymore interfaced with Yambo, and we are working together with abinit developers to restore compatibility.
They can be generated in quantum espresso format by using the Don Hamann code that can be downloaded here: http://www.mat-simresearch.com/

Best,

Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/
User avatar
Daniele Varsano
 
Posts: 1985
Joined: Tue Mar 17, 2009 2:23 pm

Re: New pseudos and coulombian cut

Postby Flex » Mon Apr 10, 2017 3:10 pm

Daniele Varsano wrote:Dear Thierry,
just a comment on what I wrote you above.
The multi-projector ONCVPSP pseudos are supported only if you perform the ground state calculations using quantum-epsresso, as in the old versions of abinit the multi projectors are not stored in the KSS file. New release of abinit instead are not anymore interfaced with Yambo, and we are working together with abinit developers to restore compatibility.
They can be generated in quantum espresso format by using the Don Hamann code that can be downloaded here: http://www.mat-simresearch.com/

Best,

Daniele


I use QE to do the ground state calculations for Yambo (as the interface with Abinit is outdated, like you said), but I want to compare with some results I obtained with Abinit alone (ground-state + GW). That is why I want to use these same pseudos. These pseudos are already available in UPF format on the pseudodojo git, FYI.

Thanks again for the helpful answers
Thierry Clette
Student at Université Libre de Bruxelles, Belgium
Flex
 
Posts: 37
Joined: Fri Mar 25, 2016 4:21 pm

Re: New pseudos and coulombian cut

Postby Flex » Thu Apr 20, 2017 3:22 pm

Daniele Varsano wrote:
Correct, but you need to:
1) Activate also the RIM, something like:
RandQpts= 1000000 # [RIM] Number of random q-points in the BZ
RandGvec= 111 RL # [RIM] Coulomb interaction RS components

Daniele


So far, I have similar results to Abinit, so the cut seems to work as intended.

I still have a small question. I don't really know if I need a lot of RandGvecs. I did some tests, from about 100 RL to 10000 RL (with 1000000 qpts, which should be enough) and it doesn't seem to change anything. (surprising, as I usually need a lot of RL for everything else, because of the empty space in the cell) So, 100 RL is probably enough ? Do I keep that value, or do I try using more as this part of the calculation takes a negligible part of the run time anyway ?
Thierry Clette
Student at Université Libre de Bruxelles, Belgium
Flex
 
Posts: 37
Joined: Fri Mar 25, 2016 4:21 pm

Re: New pseudos and coulombian cut

Postby Daniele Varsano » Fri Apr 21, 2017 7:37 am

Dear Thierry,
Usually, RandGvecs=1, e.g. treating with Monte Carlo the integral for the first Brillouin zone is enough, in some cases, you need some more, in the range of tenth, 100 maximum, it depends on the geometry of your Bz. In general, you need a lot of Gs, because you need a lot of plane waves to describe your wavefunctions, always true in presence of vacuum. Here you are not summing plane waves, by deciding which Coulomb integral to treat via Monte Carlo integration and which one in the standard way: for large G and also not so large the standard integrals are accurate so you should not see any difference as you are observing.

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/
User avatar
Daniele Varsano
 
Posts: 1985
Joined: Tue Mar 17, 2009 2:23 pm

Previous

Return to Miscellaneous

Who is online

Users browsing this forum: No registered users and 1 guest

cron