[Pw_forum] vdw interaction and pseudopotentials

[Lorenzo Paulatto]

On Saturday, October 15, 2016 12:02:51 PM CEST Zara Nosh wrote:
> Using vdw-df we should use rev-pbe pseudopotential but
> it seems that my results from blyp+vdw-df is better that revpbe+vdw-df.
> 
> Can we use blyp pseudopotentials with vdw-df?

Dear Zara,
Technically you can, but then you'll have to justify it to the referee of your 
paper who may not be too keen to accept it.

Vdw-df uses the revPBE formulation for the XC functionals, if you use anything 
else just because it fits better with what you wish to obtain, it is just 
because of a fortunate cancellation of errors: you're not doing an ab-initio 
calculation anymore.

Some amount of arbitrariness, or non-ab-initio fitting, may be inevitable even 
in a DFT calculation, some examples:
1. choosing the experimental lattice parameter instead of the optimized one; 
2. fitting the value of U in LDA+U, instead of computing it abinitio
3. fitting the fraction of the Fock exchange  in a Hybrid functional 
calculation
4. using a pseudopotential for a different flavour of the XC functional than the 
one used in you calculation

My personal opinion is:
1. acceptable if the optimized value gives completely off results, and using a 
more advanced method is not feasible
2. acceptable if computing the value of U abinitio is impossible (e.g. too 
expensive)
3. not acceptable, unless you're using HSE, in which case I have no idea
4. not acceptable, because as doing thigs properly is no more expensive and 
because I cannot find any plausible justification for it

HTH

-- 
Dr. Lorenzo Paulatto 
IdR @ IMPMC -- CNRS & Université Paris 6
phone: +33 (0)1 44275 084 / skype: paulatz
www:   http://www-int.impmc.upmc.fr/~paulatto/
mail:  23-24/4é16 Boîte courrier 115, 4 place Jussieu 75252 Paris Cédex 05