[CPMD-list] highest levels of theory

[Juerg Hutter]

Hi

> Dear list subscribers,
>
> What might be regarded as the highest
> levels of theory available in cpmd,

The level of theory in CPMD is always Kohn-Sham DFT.
Within this you can have different XC functionals.
Once you have specified the functional you have
the option to specify the basis (through the cutoff)
and choose the pseudopotentials.

The theory is defined by the functional and with
cutoff and pseudopotential you can "adjust" the
accuracy within the given theory.


> double numerical with polarization or better,
> particularly that include functionals
> appropriate for estimating dispersion forces -
> such as pbe?
>

dispersion forces and DFT is a difficult combination.
The best you can hope for is a decent description on
intermediate distances where there is still charge
overlap. Which functionals give good results, depends
on the system and probably also on your expectation
on the accuracy.


> My initial calculations on a unit cell
> of N-methylacetamide using the default
> lda correlation (PZ) with pbe (from
> contrib/pseudo/[H,C,N]_MT_PBE.psp)
> yielded very accurate frequencies for
> internal modes, with fairly accurate
> displacements (INS intensities),
> but fairly poor frequencies for interactions.
> My calculations using dnp/gga/pbe with DMol3
> have done better on all accounts except for INS intensities.
>

Using the same functional (=level of theory) one would expect
(almost) identical results between different codes. If this
is not the case you have chosen a non-sufficient basis/pseudopotential
in at least one of the codes.


> Could someone recommend a good
> combination for correlation and
> gradient correction?
> GRADIENT CORRECTION [GGAX?,PBEX?, REVPBEX?] [GGAC?,PBEC?,REVPBEC?]?
>
ALWAYS use the FUNCTIONAL keyword within the &DFT section.
The other possibilities should only used for non-standard functionals
and experts.

regards

Juerg Hutter

> --
> Dr. Robert Williams
> Dept. of Biomedical Informatics
> Uniformed Services University
> 301-295-3568
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://cpmd.org/mailman/listinfo/cpmd-list
>