The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Hi,
There is actualy a single paper explaining this kind of preconditioner: Ho et al PRB vol26 nb6 p4260 (1982).
This paper displays the LDA dielectric matrix calculations and application on the ouptut potential of the SCF cycle. Unfortunately it is not very detailed. you will just get the basic formulation. 
The RPA formulation is almost equivalent to the LDA one. Because it totaly neglects the exchange correlation part of the dielectric matrix it is much more efficient in cells containing vacuum. 

Note that in fact neither of this approach is implemented within abinit. Instead Pr. Gonze has derived a computationaly effective method (adding parameters diegap and dielam) which is the one programmed in Abinit. Unfortunately he has never find enough time to publish this.  A paper about that is in preparation but in a very (VERY) alpha state. I fear It wont be of any help for you at it's present state. 
Nevertheless feel free to ask any question you have after reading Ho's paper. 

Best regards

PMA

On 11/23/06, Jin Zhang < jin.zhang.pku@gmail.com> wrote:

Hi, Matthieu,

Would you mind giving out some relative papers concerning rpa
preconditioner? I was aware of the importance of a preconditioner, but
not sure about why. IMHO, the preconditioner only accelerates
convergence but has no effect on the final GS. Your suggestion may
solve the puzzle in my head. And I really want to dig more on this
thread.
Thanks!

Best,
Jin Zhang

--
Jin Zhang
Dep. of Physics, Peking University
Beijing, P.R.China
Phone: 86-10-6275-3190

-- 
Pierre-Matthieu Anglade