[CPMD-list] trivial tests

[Ari P Seitsonen]

Dear Alessio,

   In addition to the omission of spin-polarisation mentioned in the 
previous answer you should notice that CPMD only uses the minimal number 
of occupied orbitals, thus in this case it can pack the electron into 
three states. The usual solutions are using the keyword 'MULTIPLICITY', or 
'FREE ENERGY FUNCTIONAL' (for fractional occupations).

   The case of atomic oxygen, however, is well known to be pathodological 
for the normal Kohn-Sham calculations, namely the description would 
require a description using multi-determinental scheme: The electron could 
be occupied as
   2s2 2p_x2 2p_y1 2p_z1
   2s2 2p_x1 2p_y2 2p_z1
   2s2 2p_x1 2p_y1 2p_z2
But only one of them can be occupied simultaneously, thus usually the 
codes either converge to one of the three solutions, never converge or 
converge to 2s2 2p_x(2/3) 2p_y(2/3) 2p_z(2/3), when using fractional 
occupations. The first of the alternatives yields the lowest energy, and 
should be used in general for the calculation of binding energies etc.

   A practical problem here is that CPMD by default starts with spherical 
densities, leading to identical initial orbitals for all the three p 
components. The best way is to include some or full randomisation of the 
initial wave functions. But please be careful in order to achieve the 
converge, you might have to mix the charge density etc.

   This is an example of
  1) how a simple-sounding system can become complicated (the problem is 
there for B, C, O and F atoms - not for nitrogen, because there the p with 
spin-up shell is full, spin-down shell is empty), and
  2) how little such pre-historical issues are known nowadays; the first 
reference I know that discusses this issue is

     Phys. Rev. Lett. 59, 1285 - 1288 (1987)
     Energies of atoms with nonspherical charge densities calculated with
         nonlocal density-functional theory
     Frank W Kutzler, G S Painter
     DOI: 10.1103/PhysRevLett.59.1285

but few people seem to know about these issues, probably because 
(luckily?) a calculation for the atomic oxygen etc are seldomly needed.

   Good luck! :)

     Greetings from sunny Paris,

        apsi

On Sun, 10 Feb 2008, Alessio Alexiadis wrote:

> Hi everybody,
>
>  Ifm new to CPMD and Ifm running some trivial tests in order to understand how it works.
>  I calculated the electronic configuration of a single oxygen atom.
>
>  &CPMD
>   PROPERTIES
>   RESTART WAVEFUNCTION LATEST
>   CONVERGENCE ORBITALS
>    1.0d-7
>  &END
>
>  &SYSTEM
>   SYMMETRY
>    1
>   ANGSTROM
>   CELL
>    8.00 1.0 1.0  0.0  0.0  0.0
>   CUTOFF
>    70.0
>  &END
>
>  &DFT
>   FUNCTIONAL LDA
>  &END
>
>  &ATOMS
>  *O_MT_LDA.psp
>   LMAX=P
>    1
>   4.000   4.000   4.000
>  &END
>
>  &PROPERTIES
>  CUBEFILE ORBITALS
>   4
>   1 2 3 4 5 4
>  &END
>
>  I was expecting to obtain 2s2 [ª«] 2p4 [ª«][ª][ª], which means four wavefunctions, but I got only three wavefunctions (one s and two p). I suppose this means that the electronic distribution obtained is 2s2 [ª«] 2p4 [ª«][ª«][], which, however, is not correct according to the Hundfs rule.
>  Can anyone tell me whatfs wrong?
>  BTW, how it is possible to calculate with CPMD the excited states of an atom?
>
>  Thanks for your attention
>  Alessio
>
>
> ---------------------------------
> Looking for last minute shopping deals?  Find them fast with Yahoo! Search.

-- 
-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
   Ari P Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
   IMPMC, CNRS & Université Pierre et Marie Curie
   Tel: +33-1-4427 7542, Fax: +33-1-4427 3785, GSM: +33-6-6736 3820