The ABINIT Users Mailing List ( CLOSED )

["Tonatiuh Rangel" <Tonatiuh.Rangel@uclouvain.be>]

Dear all,

As Xavier Gonze mentioned this error was affecting Wannier90 and GW
calculations when using the PAW formalism.

Moreover, other post-processing programs such as optics_pas were affected.
The KSS files were not writen correctly, so other post-processing programs
that used KSS files as a starting point would also be affected..

This is just for the PAW formalism. Norm-conserving pseudo-potentials do
not present this error.
And it has been corrected..

Best
Tonatiuh Rangel


> Dear all,
>
> Some more information about the bug in GW ...
> (1) It is ONLY for the PAW case. Sorry for not mentioning it...
> (2) If the typat list was  2 2 2 ... 1 1 1  ...    (instead of the
> ordered 1 1 1 .... 2 2 2 ...), this bug was indeed occuring.
> (3) This affects also the use of the Wannier90 utility from ABINIT.
>
> The bug is at lines 388 and 389 of src/65nonlocal/ctoctij.F90 , in
> ABINITv5.8.4 .
> These lines
>
>      arg=two_pi*(kpoint(1)*xred(1,iatm)+kpoint(2)*xred(2,iatm)
> +kpoint(3)*xred(3,iatm))
>      phkxred(1,ia)=cos(arg);phkxred(2,ia)=sin(arg)
>
> should become
>
>      arg=two_pi*(kpoint(1)*xred(1,ia)+kpoint(2)*xred(2,ia)
> +kpoint(3)*xred(3,ia))
>      phkxred(1,iatm)=cos(arg);phkxred(2,iatm)=sin(arg)
>
> Hoping it helps,
> Xavier
>
>
>
>