Use of SCAN and rVV10 functionals in VASp 5.4.4

[jcconesa]

Hello,
The recently released 5.4.4 version of VASP is said to allow the use of SCAN (metaGGA) and rVV10 (van der Waals) functionals. Please inform on the keywords allowing their use, and on any additional keyword values which may be relevant for the correct operation of these functionals in VASP.
Regards,
José C. Conesa

[admin]

Published in PRL 115 036402 (2015)

[jcconesa]

I do not see in that article, nor in its Supplementary information, which are the VASP keywords (and their values) that must be used to make SACN and rVV10 calculations.
And the manual or the wiki do not yet include them.
Please help

[admin]

METAGGA = SCAN
https://cms.mpi.univie.ac.at/wiki/index.php/METAGGA
The rVV10 is ONLY switched on if LUSE_VDW is set. if (and only if) this is the case, rVV10 is used automatically

[rgc]

METAGGA = SCAN
https://cms.mpi.univie.ac.at/wiki/index.php/METAGGA
The rVV10 is permanently switched on.

Is there a way then of doing a SCAN calculation without the rVV10 correction?

[msd30]

Hi

As far as I can see in v 5.4.4 it is necessary to specify LUSE_VDW=.TRUE. along with METAGGA=SCAN in the INCAR file to obtain the SCAN+rVV10 combination.

Without specifying LUSE_VDW=.TRUE. I do not receive the line "Total rVV10 correction in eV: ..." in the standard output, which I assume means that the rVV10 routines are not entered automatically without this flag.

Also, as far as I can see, the b parameter for rVV10 is fixed to 6.3 in vdw_nl.F, whereas the recommended value for SCAN+rVV10 is that b = 15.7 (Peng et al., Phys Rev X 6 (2016) 041005). It would be good to know whether this is indeed the case, since work published using the SCAN+rVV10 functional in VASP would then not be consistent with the original published SCAN+rVV10 functional.

Matthew

[ehermes]

The rVV10 is permanently switched on.

I do not believe this is the case. According to Haowei Peng, one must set "LUSE_VDW = .TRUE." and "BPARAM = 15.7" to perform a SCAN-rVV10 calculation. Indeed, I have tried calculating the lattice constant of Pd using SCAN with and without these arguments, and the results indicate that without those keywords the rVV10 calculation is not performed based on how they compare to the benchmark values from the original SCAN-rVV10 paper. Moreover, unless "LUSE_VDW = .TRUE." is specified, the results are not affected by changing the "BPARAM" keyword.

[admin]

This is the correct answer, see here.