which version of paw potential used in related vdw-DF calculation? LDA or GGA (PW91, PBE)

Message

njuxyh · #1 Post by **njuxyh** » Fri Nov 11, 2011 9:27 am

hi :

i have another newbie question about the PAW potential used in related vdw-DF, in the reference PRB 83,195131 (2011), I have a look at the vdw-DF frame, the EXC including three parts, including LDA correlation part and GGA exchange part. of cource, i know i should set GGA=OR or B0, and so on. then The XC part will adjust according with vdw-DF frame based the given PAW potential. but i stiil wondering which version of PAW should be used,( LDA OR GGA )? which is better or fast for practical calculation?

THANK YOU !

kelum · #2 Post by **kelum** » Fri Nov 11, 2011 1:16 pm

In principle both LDA or PBE POTCARs should work. We tend to use PBE PAWs because one doesn't have to change them when comparing to a PBE calculation. We did some comparison for graphite and found basically no difference between LDA and PBE POTCARs both hard and normal ones.

I don't think the speed should differ.

See also the second point in the "Important remarks" part of the vdW-DF VASP manual, the vdw correction is calculated on the valence electron density (+core corrections) and will get more accurate if one uses hard PAWs or PAWs with more electrons. However, the results usually don't differ significantly between normal PAWs and hard PAWs. But you can use the hard/more electron PAWs to check that the accuracy is all right with the normal PAWs.