[Pw_forum] dos

Gabriele Sclauzero sclauzer at sissa.it
Tue May 5 15:42:10 CEST 2009



Jonas Baltrusaitis wrote:
> Gabriele,
> 
> you mean physically sum up all, say, C2s PDOS? There is no averaging or similar task
involved?

It depends what you want to look at... for instance, if I want to find the energy position 
of the \pi states of a CO molecule (aligned along z) I could sum up the s and p_z PDOS of 
C and O atoms.
Since the units of PDOS are electrons/ev (apart from some spilling charge due to 
incompleteness of atomic basis in the solid), I could integrate the summed PDOS and find 
that each \pi peak corresponds to an integer number of electrons (which is given by the 
degeneracy of the state).
Averaging does not make much sense in this case, since C and O are obviously not 
equivalent. At the moment I cannot figure out a case where it could have.
Also averaging PDOS of equivalent atoms is not useful, since PDOS on all equivalent atoms 
should be equal.
I don't know what else you were meaning by "averaging", sorry.

GS

> 
> 
> --- On Tue, 5/5/09, Gabriele Sclauzero <sclauzer at sissa.it> wrote:
> 
>> From: Gabriele Sclauzero <sclauzer at sissa.it>
>> Subject: Re: [Pw_forum] dos
>> To: "PWSCF Forum" <pw_forum at pwscf.org>
>> Date: Tuesday, May 5, 2009, 12:26 AM
>> Stefano Baroni wrote:
>>> Jonas: I am afraid that what you want to obtain does
>> not make much 
>>> sense. The angular character of any molecular (or
>> Bloch) state depends 
>>> on the origin of your reference system. What is
>> "s-like" with respect to 
>>> one point has an infinite number of angular components
>> with respect to a 
>>> different point. Are you sure you want to add
>> "what is s-like in Iowa" 
>>> with "what is s-like in Italy" ??? ;-)
>>> Stefano
>> I agree with what you say, Stefano, about taking the
>> expectation value of the angular 
>> momentum in a periodic system, but summing the PDOS of
>> selected atoms can lead to 
>> meaningful and sometimes very useful results, in some
>> cases.
>> I usually see the PDOS as an indicator of how much the
>> Bloch states retain a kind of 
>> "atomic character" in the solid (so that in this
>> case summing PDOS of equivalent atoms is 
>> the more meaningful thing, as Lex pointed out), but also as
>> a way to identify the symmetry 
>> of states.
>> If you place a molecule, a nanowire or maybe even a surface
>> with slab geometry in a wise 
>> manner in your (super-)cell you can get many information
>> from the PDOS (summed or not).
>> Obviously, it does not make much sense to sum the s,p,d
>> PDOS of all atoms in the system in 
>> general (but in some cases it may), so that's why you
>> cannot find it directly in the 
>> output from projwfc.
>> It's your task to select over which atoms you can do a
>> sum (or an average) of the PDOS 
>> (and also placing the atoms in a smart way, first). I think
>> you can write a script in 
>> awk/perl/whatever, or even a small fortran program, which
>> reads PDOS from individual files 
>> associated to the selected atoms and sums the corresponding
>> rows.
>>
>> Gabriele
>>
>>> On May 4, 2009, at 5:51 PM, Jonas Baltrusaitis wrote:
>>>
>>>> I just successfully calculated DOS and PDOS of a
>> very large system. 
>>>> One thing that confuses me is the presence of s, p
>> components to PDOS 
>>>> for every atom, but not summed (or averaged) s and
>> p contribution of 
>>>> all system. I would be happy to find in my output
>> total DOS, PDOS s 
>>>> character and PDOS p character. Is there a way to
>> get that instead of 
>>>> contribution per every atom? Or how do I proceed
>> to get s and p PDOS 
>>>> for all system by manually processing all of my 76
>> atoms PDOS outputs?..
>>>> thanks
>>>>
>>>> Jonas Baltrusaitis
>>>> University of Iowa

-- 


o ------------------------------------------------ o
| Gabriele Sclauzero, PhD Student                  |
| c/o:   SISSA & CNR-INFM Democritos,              |
|        via Beirut 2-4, 34014 Trieste (Italy)     |
| email: sclauzer at sissa.it                         |
| phone: +39 040 3787 511                          |
| skype: gurlonotturno                             |
o ------------------------------------------------ o



More information about the users mailing list