[QE-developers] Little error in vexx_k?

[Paolo Giannozzi]

I don't think there is any error, just some confusion due to the rather
complex implementation of parallelization over pairs of bands in exact
exchange. The second dimension of arrays "psi" and "hpsi" does not matter
in practice, as long as nothing goes out of bounds. Unused variables "fac"
are the consequences of changes made to improve openMP parallelization.

I am not saying that I am happy with the current status of the
exact-exchange routines. quite the opposite: I am VERY unhappy and I think
that some serious cleaning is badly needed (also in conjunction with the
forthcoming inclusion of the new code with localization)

Paolo

On Fri, Dec 14, 2018 at 4:26 PM Lucas <lucas.baguet at upmc.fr> wrote:

> Dear QE developers,
>
> I have a question concerning a routine in the "exx" library.
>
> In "vexx_k", we want to apply the vexx potential to "m" vectors ("m"
> being an input parameter). However, the arrays "psi" and "hpsi" are
> declared with "max_ibands" as second dimension. "max_ibands" is not used
> elsewhere in the routine, but "m" is. Note also that in "vexx_gamma" the
> second dimension of "psi" and "hpsi" is "m".
>
> Is there some reason for keeping "max_ibands" in that case?
>
> Also, in both "vexx_k" and "vexx_gamma", a local array "fac" is declared
> and allocated but does not seem to be used.
>
> Sincerely,
> Lucas Baguet
>
> PS : I noticed that point while working on the ACFDT code with Maria
> Hellgren and Lorenzo Paulatto at the IMPMC, Sorbonne Université, Paris.
> _______________________________________________
> developers mailing list
> developers at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/developers
>


-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20181220/c8e05df7/attachment.html>