[Wannier] Eq. (B1) not satisfied in kmesh_get (1)
Jian-Xin Zhu
jxzhu at lanl.gov
Fri Nov 19 02:55:02 CET 2010
Dear Jonathan,
If it goes this way, I may avoid the error by using rational fractions
in all three directions, like 1/20 1/20 1/10 (if allowed for a
specific crystal structure).
Does it make sense?
Thanks,
Jianxin
On Nov 18, 2010, at 5:14 PM, Jonathan Yates wrote:
>
> On 18 Nov 2010, at 23:47, Jian-Xin Zhu wrote:
>> One remark: The above listed k-values are generated from the
>> wien2wannier interface.
>> I have a strange observation. For the number of k-points of like 6
>> 6 6 (a fcc structure), the error is encountered.
>> For the number of k-points like 8 8 8, the error does not appear.
>> However, if I use even larger number of k-points,
>> e.g., 12 12 12, the error comes up again. It seems there is no
>> guarantee to resolve the problem by increasing the
>> number of k-points (Of course, the calculation also becomes more
>> expensive with large number of k-points).
>
> There is simple explanation:
> 1/8 can be exactly represented with 3decimal places of precision.
> 1/6 and 1/12 cannot be precisely expressed with a finite number of
> decimal places.
> So the fact that you see the error with grids of size 6 and 12 but
> not 8 shows us that it is a precision error.
>
> So either you increase the precision of the kpoints. Or decrease the
> value of kmesh_tol. They will both achieve the same end.
>
> Jonathan
>
>
>
>
>
> --
> Department of Materials, University of Oxford, Parks Road, Oxford,
> OX1 3PH, UK
> tel: +44 (0)1865 612797 http://users.ox.ac.uk/
> ~oums0549/
>
> _______________________________________________
> Wannier mailing list
> Wannier at quantum-espresso.org
> http://www.democritos.it/mailman/listinfo/wannier
--
###############################
Jian-Xin Zhu, Ph.D
Theorertical Division, MS B262
Los Alamos National Laboratory
Los Alamos, NM 87545
Phone: (505) 667 2363
Fax: (505) 665 4063
Emai: jxzhu at lanl.gov
Email (backup): physjxzhu at gmail.com
URL: http://theory.lanl.gov
###############################
More information about the Wannier
mailing list