Using meta-GGA in vasp

Message

a.soon · #1 Post by **a.soon** » Sun Dec 05, 2010 11:29 pm

Dear vasp users,

I was wondering if someone could kindly provide the keywords for specifying the use of meta-GGAs in a vasp calculation?

I understand that it is now possible to run self-consistent meta-GGAs (TPSS and revTPSS) and the MBJ meta-GGA.

Thanks in advance.

cheers, Aloysius

#2 Post by **admin** » Thu Dec 30, 2010 1:47 pm

these features are not supported officially at the moment, you will need a new set of pseudopotentials which have not been released yet

WolverBean · #3 Post by **WolverBean** » Thu Oct 11, 2012 12:24 am

With the release of a new set of pseudopotentials for VASP 5.2, are meta-GGA's now supported? If so, what is the keyword for specifying their use?

#4 Post by **admin** » Fri Oct 12, 2012 12:26 pm

New potpaw.52 potentials fully support meta-GGA calculations. The calculation is specified with METAGGA = XXX (XXX = M06L, TPSS, RTPSS, or MBJ). First three are SCF, the MBJ is non-SCF.

askhetan · #5 Post by **askhetan** » Wed Jan 09, 2013 8:49 pm

Dear Admin,
In general, meta-GGAs are quite good and perhaps better than GGAfunctionals like RPBE, etc. Please let me know a simple thing. For my calculations on CeO2 (with problematic f electrons) can I use meta-GGA+U just like normal GGA+U that I have beenusing. DOest it make any sense?

Or can I expect reasonable accuracy for CeO2 with just meta-GGA (without U).

Also, the manual says that for safe calculations LMAXTAU=2+L. So in this case of f electrons, can I use LMAXTAU=8 for safety ?

Thanks in advance, hoping for a reply!

#6 Post by **admin** » Fri Jan 11, 2013 2:20 pm

meta-GGA does not treat the problem of the strong correlation. For this LDA(GGA)+U method is used. meta-GGA can be combined with GGA+U.

WolverBean · #7 Post by **WolverBean** » Sat Feb 02, 2013 3:33 am

askhetan,

I can add that in my recent experience comparing RPBE+U calculations to M06L calculations on systems with partially occupied d orbitals, a +U correction was not needed to get good results with M06L. I can't speak to whether you'll find the same to be true with occupied f orbitals, but it seems possible.
Yes, use LMAXTAU=8.

askhetan · #8 Post by **askhetan** » Mon Feb 04, 2013 11:55 am

Thank You! I will try out comparing the properties with M06L then.

FyneMan · #9 Post by **FyneMan** » Wed May 29, 2013 1:16 pm

[quote=" XXX (XXX = M06L, TPSS, RTPSS, or MBJ). First three are SCF, the MBJ is non-SCF.[/quote"]

Dear Admin

I use MBJ to calculate the band gap of some semiconductor materials. First, I do a scf calculation using PBE functional. Then I do a MBJ calculation from the wavefunction of the previous step with a fixed value of the CMBJ-tag. It is difficult to get convergent in the second step. I set ISMEAR = 0, SIGMA = 0.1, ALGO = FAST. Could you give me some suggestions to accelerate the convergence?
Thanks

FyneMan