Re: [Spglib-users] symmetrizing positions/cell vectors without altering cell?

[Noam B. <noa...@nr...>]

> On Dec 14, 2017, at 7:31 PM, Atsushi Togo <atz...@gm...> wrote:
> 
> No.
> 
> Cell vectors follow the crystallographic point group (multiplied with
> translation group). But if I remember correctly, simple application of
> these symmetry operations similarly to the points of atoms didn't work
> well. So my strategy is that using the constraints of Bravais lattice
> for angles and equivalent basis vector lengths to symmetrize the basis
> vectors. For distorted basis vectors, it's not uniquely defined the
> directions of basis vectors in Cartesian coordinates, I employed a
> strategy to align those basis vectors to the Cartesian axes as shown
> in the spglib documentation. If you want to rotate basis vectors to be
> along some directions, you can do rotate back afterwards using the
> rotation matrix calculated from the initial and final basis vectors
> and the transformation matrix given from spglib. In summary, to
> symmetrize basis vectors, there is large freedom to make it, and it
> can be dependent on users' will, so I don't want to provide a function
> that does too much.

That sounds reasonable, except that I can’t figure out how to decompose
the transformation from my original vectors to the symmetrized vectors into
a supercell+deformation part and a rotation.  standardize_cell just returns
a set of final lattice vectors. Is that separation already someplace in the code, 
or do I need to decompose the transformation matrix myself somehow?  

								Noam