[Pw_forum] PAW in 4.0CVS3

Lorenzo Paulatto paulatto at sissa.it
Thu Apr 17 10:12:12 CEST 2008


On Gio, Aprile 17, 2008 02:57, Chao Cao wrote:
>    Is there any major change or bug fix for PAW method between CVS2 and
>  CVS3?

As far as I remember, there should be exactly zero of them.

> I used exactly the same set of PP-generation input and PW input,
>  but somehow got very different result in total energy and force (differs
>  by several eV).

I think you should regenerate your pseudopotentials files, as I have
revised the format, maybe the new version is misreading the old files.

> [cut]
> Since I set conv_thr to be 1.0d-7,
>  a 0.0004 Ry difference in total energy looks really big to me...

It definitely is, but I never had such a problem. Unless you have done
some serious mistake in you input (i.e. cutoff far too low) it should not
happen.

There is a very stupid bug in the CVS3 version, it may cause random data
to be used for augmentation charge. If you want to fix it by hand, you
have to actually set to zero the "zeros" array in read_nonlocal routine,
in Modules/read_upf_v2.f90.

I hope whis will fix your problems, but if they persist, please report.

bye

-- 
Lorenzo Paulatto
SISSA  &  DEMOCRITOS (Trieste)
+39 040 3787 511
http://people.sissa.it/~paulatto/


----------------------------------------------------------------
  SISSA Webmail https://webmail.sissa.it/
  Powered by SquirrelMail http://www.squirrelmail.org/




More information about the users mailing list