[Python] How to apply rotations to structured numpy arrays?

I'm using structured arrays to store atoms data produced by LAMMPS (I'm using a structured array that follows its format). I need to rotate the positions:

import numpy as np

transform = np.array([[1, 0, 0], [0, 1, 0], [0, 0, 1]], dtype=np.float64)
dtype = np.dtype([("x", np.float64), ("y", np.float64), ("z", np.float64)])

atoms = np.array(
[
(0.0, 0.0, 0.0),
(1.0, 0.0, 0.0),
(0.0, 1.0, 0.0),
(1.0, 1.0, 1.0),
],
dtype=dtype,
)

atoms[["x", "y", "z"]] = atoms[["x", "y", "z"]] @ transform.T



But this produces:

Traceback (most recent call last):
File "c:\Users\acgc99\Desktop\rotation.py", line 16, in <module>
atoms[["x", "y", "z"]] = atoms[["x", "y", "z"]] @ transform.T
~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~
numpy._core._exceptions._UFuncNoLoopError: ufunc 'matmul' did not contain a loop with signature matching types (dtype([('x', '<f8'), ('y', '<f8'), ('z', '<f8')]), dtype('float64')) -> None


I can convert to unstructured arrays, but I guess that doing that change multiple times is not efficient when working with tens of millions of atoms.

Continue reading...