The ABINIT Users Mailing List ( CLOSED )

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

On Wed, Apr 22, 2009 at 4:33 PM, florine bonald
<florine.bonald@gmail.com> wrote:
> Hi,
>
> Reading the answer of this question and the question I had on
> pseudopotentials a couple of days ago, it is obvious that changing ixc alone
> in the input file does not have much effect on the results.

Usually the results don't change tremendously when you use ixc.
However it always results in some changes. And those changes can be
stronger than what you get with a new pseudo based on an other XC
functional. I don't think there is any general rule.

> Therefore, for
> any new amount of ixc in the input, we have to regenerate the
> pseudopotential, is that right?

It's most coherent: regenerating the pseudo you use a single XC
approximation instead of two. And, as mentionned by Andrew you are
sure that at least for one atomic configuration you get the exact same
energy as an all electron calculation. But it is not very clear if
using a single XC functional is better than mixing two since anyway no
density based XC functional is perfect.

>
> Not scaping from the difficulty, but obtaining a well-tested pseudo is a
> demanding job. Please let me know if there is another way to obtain the TM
> pseudos for different XC.

>
> Regards,
> Florin
>
>



-- 
Pierre-Matthieu Anglade