Transferability of van der Waals parameters across schemes

[hipertrofia]

I am running molecular dynamics with the RPBE functional (GGA = RP) and have been trying to incorporate van der Waals corrections. The damping parameters for this functional are available for the D3 correction (IVDW = 11) but not for the Tkatchenko-Scheffler (IVDW = 2) scheme. However, using the D3 scheme makes my calculations about 5 times as expensive as without the dispersion correction - this is completely unaffordable.

I have previously run Tkatchenko-Scheffler corrected PBE simulations which had about the same CPU cost as the uncorrected ones. Since by using TS with RPBE I need to provide my own VDW_SR parameter, I was wondering what are issues associated with taking the PBE one for use with RPBE. Or can someone provide a reference of optimized parameters for RPBE + TS? Also, would using D3 with Becke-Johnson damping (IVDW = 12) improve the efficiency of the approach?

Any directions are much appreciated.

[admin]

You need to adapt parameters for the RPBE functional.
It is done via the error minimization for the set S22
(cf. A. Tkachenko et al. J. Chem. Phys. 131, 094106 (2009)).
The adaptation was already done for many functionals
(cf. A. Tkachenko et al. J. Chem. Theory Comput. 7 3944-3951 (2011)).
but not yet for RPBE.
PS The slowness of the D3 code is known issue.

[hipertrofia]

Thanks for the reply. Since I couldn't find values in the literature did the optimization of VDW_SR myself for RPBE (VDW_SR = 0.59) and several other GGAs. Here is the reference: https://arxiv.org/abs/1704.00761

[admin]

Thank you.