Re: [Spglib-users] Bug in symmetry detection in spglib 1.15

[Atsushi T. <atz...@gm...>]

Hi Lorenzo,
(I forgot to reply to spglib-users.)

Thanks for your report. This issue will be easier to discuss on spglib
github issues, https://github.com/spglib/spglib/issues, so could you
open an issue on this? In addition, I want to ask you not to use ASE
but use the tuple format written here,
https://spglib.github.io/spglib/python-spglib.html#crystal-structure-cell.

Togo

On Sat, Jun 6, 2020 at 1:25 AM Lorenzo Monacelli
<mes...@gm...> wrote:
>
> Dear developers,
>
> I think I found a bug that is affecting the newest version of spglib
> (version 1.14 works), or at least the python interface.
>
> When I ask the symmetry group of a structure by with a python script, by
> passing a threshold, it correctly identify the group.
>
> However, when I ask the symmetry operations, it seems to ignore the
> threshold parameters.
>
> I attach an example that works properly with spglib==1.14.1 but not with
> spglib==1.15.1.
>
> The example requires ASE installed to read the ".cif" file. It is a
> slightly displaced C2/c group structure.
>
> If I ask the symmetry group without specifying a threshold, it correctly
> says it is P1 (1), while if I use a small threshold (0.05) it recognizes
> the C2/c (4) group.
>
> Then I ask the code the symmetries and the rotations, by specifying the
> same threshold that allows to recognize the C2/c group, but it returns
> just the identity matrix.
>
> As I told you, if I use an older version of spglib, the code works properly.
>
>
> Is it a bug? will it be solved in the next release?
>
> Thanks for your attention,
>
> Best regards,
>
> Lorenzo Monacelli
>
> _______________________________________________
> Spglib-users mailing list
> Spg...@li...
> https://lists.sourceforge.net/lists/listinfo/spglib-users



-- 
Atsushi Togo