[Pw_forum] [***SPAM*** Score/Req: 10.00/5.0] Re: [***SPAM*** Score/Req: 10.00/5.0] Re: (no subject)

[Guido Fratesi]

> But when we divide the unit cell into 2 parts
> and we use projwfc.x for calculation of local dos, the sum of two local
> dos columns is smaller than total dos.

You are probably using USPP and missing the augmentation charge 
contribution.

> I will be very grateful ifyou can give me a suggestion or reference for
> more information about the origin of the difference between result of
> local dos and summation of projected dos in the same volume.

These are two different concepts (below, psi_nk=bands; phi=atomic wfc):

PDOS_nlm: sum_nk weight(k) occupation(n,k) | <psi_n(k)|phi_nlm> |^2

LDOS_vol: sum_nk weight(k) occupation(n,k) \int_vol d3r |psi_n(k)(r)|^2

Try to figure out them, there's no reason why they would be the same. 
There's no "same volume" for PDOS and LDOS. Only the sum of all PDOS_nlm 
should equal the total DOS. Idem, the LDOS in the cell volume should be 
the total DOS but the augmentation contribution to |psi(r)| was not 
implemented.

> PS: I want to compare total and adsorbate-projected density of states,
> which one is correct method; summation over all projected dos on
> adsorbate atoms or calculation of local dos for the box which contains
> all of the atoms of adsorbate?

Keeping in mind that arbitrariness is present in both methods, and 
qualitative questions can be answered, I would use the PDOS. Try answer 
the questions: "what is THE box which contains all of the atoms of the 
adsorbate?" Can you? Or worse... "Do electrons belong to this atom?"

Guido

PS there's extensive literature on the subject of partitioning charge:
very recently, http://link.aps.org/doi/10.1103/PhysRevLett.108.166403

-- 
Guido Fratesi

Dipartimento di Scienza dei Materiali
Universita` degli Studi di Milano-Bicocca
via Cozzi 53, 20125 Milano, Italy