This is under your own responsibility.
First choose the functional in DFT, LDA or GGA+U,
but +U is required to catch, even partially,
what Federico explained in the previous email about lanthanide.
I suppose you possess already a coherent pseudopotential for your
lanthanide(s) (?)
Then you do a convergence study => energy shift, psp cutoff(s),
meshcutoff, +U parameters are studied up to obtain something close to
your PW results.
The coherence of this procedure depends on what you need/want to show
with Siesta, I mean
if it is structural parameters, a band structure, it should be doable
with an agreement below
a % and/or a few tenths of eV.
Le 04/04/2014 20:11, Wei Hu a écrit :
How about the atomic orbitals describe these systems compared with plane wave
basis set?
