biotite 1.5.0__cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl

biotite/structure/box.py

@@ -0,0 +1,724 @@
+# This source code is part of the Biotite package and is distributed
+# under the 3-Clause BSD License. Please see 'LICENSE.rst' for further
+# information.
+
+"""
+Functions related to working with the simulation box or unit cell
+of a structure.
+"""
+
+__name__ = "biotite.structure"
+__author__ = "Patrick Kunzmann"
+__all__ = [
+    "space_group_transforms",
+    "vectors_from_unitcell",
+    "unitcell_from_vectors",
+    "box_volume",
+    "repeat_box",
+    "repeat_box_coord",
+    "move_inside_box",
+    "remove_pbc",
+    "remove_pbc_from_coord",
+    "coord_to_fraction",
+    "fraction_to_coord",
+    "is_orthogonal",
+]
+
+import functools
+import json
+from numbers import Integral
+from pathlib import Path
+import numpy as np
+import numpy.linalg as linalg
+from biotite.structure.atoms import repeat
+from biotite.structure.chains import get_chain_masks, get_chain_starts
+from biotite.structure.error import BadStructureError
+from biotite.structure.molecules import get_molecule_masks
+from biotite.structure.transform import AffineTransformation
+from biotite.structure.util import vector_dot
+
+
+def space_group_transforms(space_group):
+    """
+    Get the coordinate transformations for a given space group.
+
+    Applying each transformation to a structure (in fractional coordinates) reproduces
+    the entire unit cell.
+
+    Parameters
+    ----------
+    space_group : str or int
+        The space group name (full *Hermann-Mauguin* symbol) or
+        *International Table*'s number.
+
+    Returns
+    -------
+    transformations : list of AffineTransformation
+        The transformations that creates the symmetric copies of a structure in a unit
+        cell of the given space group.
+        Note that the transformations need to be applied to coordinates in fractions
+        of the unit cell and also return fractional coordinates, when applied.
+
+    See Also
+    --------
+    coord_to_fraction : Used to convert to fractional coordinates.
+    fraction_to_coord : Used to convert back to Cartesian coordinates.
+
+    Examples
+    --------
+
+    >>> transforms = space_group_transforms("P 21 21 21")
+    >>> for transform in transforms:
+    ...     print(transform.rotation)
+    ...     print(transform.target_translation)
+    ...     print()
+    [[[1. 0. 0.]
+      [0. 1. 0.]
+      [0. 0. 1.]]]
+    [[0. 0. 0.]]
+    <BLANKLINE>
+    [[[-1.  0.  0.]
+      [ 0. -1.  0.]
+      [ 0.  0.  1.]]]
+    [[0.5 0.0 0.5]]
+    <BLANKLINE>
+    [[[-1.  0.  0.]
+      [ 0.  1.  0.]
+      [ 0.  0. -1.]]]
+    [[0.0 0.5 0.5]]
+    <BLANKLINE>
+    [[[ 1.  0.  0.]
+      [ 0. -1.  0.]
+      [ 0.  0. -1.]]]
+    [[0.5 0.5 0.0]]
+    <BLANKLINE>
+
+    Reproduce the unit cell for some coordinates (in this case only one atom).
+
+    >>> asym_coord = np.array([[1.0, 2.0, 3.0]])
+    >>> box = np.eye(3) * 10
+    >>> transforms = space_group_transforms("P 21 21 21")
+    >>> # Apply the transformations to fractional coordinates of the asymmetric unit
+    >>> unit_cell = np.concatenate(
+    ...     [
+    ...         fraction_to_coord(transform.apply(coord_to_fraction(asym_coord, box)), box)
+    ...         for transform in transforms
+    ...     ]
+    ... )
+    >>> print(unit_cell)
+    [[ 1.  2.  3.]
+     [ 4. -2.  8.]
+     [-1.  7.  2.]
+     [ 6.  3. -3.]]
+    """
+    transformation_data = _get_transformation_data()
+
+    if isinstance(space_group, str):
+        try:
+            space_group_index = transformation_data["group_names"][space_group]
+        except KeyError:
+            raise ValueError(f"Space group '{space_group}' does not exist")
+    else:
+        try:
+            space_group_index = transformation_data["group_numbers"][str(space_group)]
+        except KeyError:
+            raise ValueError(f"Space group number {space_group} does not exist")
+
+    space_group = transformation_data["space_groups"][space_group_index]
+    transformations = []
+    for transformation_index in space_group:
+        matrix = np.zeros((3, 3), dtype=np.float32)
+        translation = np.zeros(3, dtype=np.float32)
+        for i, part_index in enumerate(
+            transformation_data["transformations"][transformation_index]
+        ):
+            part = transformation_data["transformation_parts"][part_index]
+            matrix[i, :] = part[:3]
+            translation[i] = part[3]
+        transformations.append(
+            AffineTransformation(
+                center_translation=np.zeros(3, dtype=np.float32),
+                rotation=matrix,
+                target_translation=translation,
+            )
+        )
+    return transformations
+
+
+def vectors_from_unitcell(len_a, len_b, len_c, alpha, beta, gamma):
+    """
+    Calculate the three vectors spanning a box from the unit cell
+    lengths and angles.
+
+    The return value of this function are the three box vectors as
+    required for the :attr:`box` attribute in atom arrays and stacks.
+
+    Parameters
+    ----------
+    len_a, len_b, len_c : float
+        The lengths of the three box/unit cell vectors *a*, *b* and *c*.
+    alpha, beta, gamma : float
+        The angles between the box vectors in radians.
+        *alpha* is the angle between *b* and *c*,
+        *beta* between *a* and *c*, *gamma* between *a* and *b*.
+
+    Returns
+    -------
+    box : ndarray, dtype=float, shape=(3,3)
+        The three box vectors.
+        The vector components are in the last dimension.
+        The value can be directly used as :attr:`box` attribute in an
+        atom array.
+
+    See Also
+    --------
+    unitcell_from_vectors : The reverse operation.
+    """
+    a_x = len_a
+    b_x = len_b * np.cos(gamma)
+    b_y = len_b * np.sin(gamma)
+    c_x = len_c * np.cos(beta)
+    c_y = len_c * (np.cos(alpha) - np.cos(beta) * np.cos(gamma)) / np.sin(gamma)
+    c_z = np.sqrt(len_c * len_c - c_x * c_x - c_y * c_y)
+    box = np.array([[a_x, 0, 0], [b_x, b_y, 0], [c_x, c_y, c_z]], dtype=np.float32)
+
+    # Fix numerical errors, as values, that are actually 0,
+    # might not be calculated as such
+    tol = 1e-4 * (len_a + len_b + len_c)
+    box[np.abs(box) < tol] = 0
+
+    return box
+
+
+def unitcell_from_vectors(box):
+    """
+    Get the unit cell lengths and angles from box vectors.
+
+    This is the reverse operation of :func:`vectors_from_unitcell()`.
+
+    Parameters
+    ----------
+    box : ndarray, shape=(3,3)
+        The box vectors.
+
+    Returns
+    -------
+    len_a, len_b, len_c : float
+        The lengths of the three box/unit cell vectors *a*, *b* and *c*.
+    alpha, beta, gamma : float
+        The angles between the box vectors in radians.
+
+    See Also
+    --------
+    vectors_from_unitcell : The reverse operation.
+    """
+    a = box[0]
+    b = box[1]
+    c = box[2]
+    len_a = linalg.norm(a)
+    len_b = linalg.norm(b)
+    len_c = linalg.norm(c)
+    alpha = np.arccos(np.dot(b, c) / (len_b * len_c))
+    beta = np.arccos(np.dot(a, c) / (len_a * len_c))
+    gamma = np.arccos(np.dot(a, b) / (len_a * len_b))
+    return len_a, len_b, len_c, alpha, beta, gamma
+
+
+def box_volume(box):
+    """
+    Get the volume of one ore multiple boxes.
+
+    Parameters
+    ----------
+    box : ndarray, shape=(3,3) or shape=(m,3,3)
+        One or multiple boxes to get the volume for.
+
+    Returns
+    -------
+    volume : float or ndarray, shape=(m,)
+        The volume(s) of the given box(es).
+    """
+    # Using the triple product
+    return np.abs(linalg.det(box))
+
+
+def repeat_box(atoms, amount=1):
+    r"""
+    Repeat the atoms in a box by duplicating and placing them in
+    adjacent boxes.
+
+    The output atom array (stack) contains the original atoms (central
+    box) and duplicates of them in the given amount of adjacent boxes.
+    The coordinates of the duplicate atoms are translated accordingly
+    by the box coordinates.
+
+    Parameters
+    ----------
+    atoms : AtomArray or AtomArrayStack
+        The atoms to be repeated.
+        If `atoms` is a :class:`AtomArrayStack`, the atoms are repeated
+        for each model, according to the box of each model.
+    amount : int, optional
+        The amount of boxes that are created in each direction of the
+        central box.
+        Hence, the total amount of boxes is
+        :math:`(1 + 2 \cdot \text{amount}) ^ 3`.
+        By default, one box is created in each direction, totalling in
+        27 boxes.
+
+    Returns
+    -------
+    repeated : AtomArray or AtomArrayStack
+        The repeated atoms.
+        Includes the original atoms (central box) in the beginning of
+        the atom array (stack).
+        If the input contains the ``sym_id`` annotation, the IDs are continued in the
+        repeated atoms, i.e. they do not start at 0 again.
+    indices : ndarray, dtype=int, shape=(n,3)
+        Indices to the atoms in the original atom array (stack).
+        Equal to
+        ``numpy.tile(np.arange(atoms.array_length()), (1 + 2 * amount) ** 3)``.
+
+    See Also
+    --------
+    repeat_box_coord : Variant that acts directly on coordinates.
+
+    Examples
+    --------
+
+    >>> array = AtomArray(length=2)
+    >>> array.coord = np.array([[1,5,3], [-1,2,5]], dtype=float)
+    >>> array.box = np.array([[10,0,0], [0,10,0], [0,0,10]], dtype=float)
+    >>> repeated, indices = repeat_box(array)
+    >>> print(repeated.coord)
+    [[  1.   5.   3.]
+     [ -1.   2.   5.]
+     [ -9.  -5.  -7.]
+     [-11.  -8.  -5.]
+     [ -9.  -5.   3.]
+     [-11.  -8.   5.]
+     [ -9.  -5.  13.]
+     [-11.  -8.  15.]
+     [ -9.   5.  -7.]
+     [-11.   2.  -5.]
+     [ -9.   5.   3.]
+     [-11.   2.   5.]
+     [ -9.   5.  13.]
+     [-11.   2.  15.]
+     [ -9.  15.  -7.]
+     [-11.  12.  -5.]
+     [ -9.  15.   3.]
+     [-11.  12.   5.]
+     [ -9.  15.  13.]
+     [-11.  12.  15.]
+     [  1.  -5.  -7.]
+     [ -1.  -8.  -5.]
+     [  1.  -5.   3.]
+     [ -1.  -8.   5.]
+     [  1.  -5.  13.]
+     [ -1.  -8.  15.]
+     [  1.   5.  -7.]
+     [ -1.   2.  -5.]
+     [  1.   5.  13.]
+     [ -1.   2.  15.]
+     [  1.  15.  -7.]
+     [ -1.  12.  -5.]
+     [  1.  15.   3.]
+     [ -1.  12.   5.]
+     [  1.  15.  13.]
+     [ -1.  12.  15.]
+     [ 11.  -5.  -7.]
+     [  9.  -8.  -5.]
+     [ 11.  -5.   3.]
+     [  9.  -8.   5.]
+     [ 11.  -5.  13.]
+     [  9.  -8.  15.]
+     [ 11.   5.  -7.]
+     [  9.   2.  -5.]
+     [ 11.   5.   3.]
+     [  9.   2.   5.]
+     [ 11.   5.  13.]
+     [  9.   2.  15.]
+     [ 11.  15.  -7.]
+     [  9.  12.  -5.]
+     [ 11.  15.   3.]
+     [  9.  12.   5.]
+     [ 11.  15.  13.]
+     [  9.  12.  15.]]
+    >>> print(indices)
+    [0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0
+     1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1]
+
+    The ``sym_id`` is continued in the repeated atoms.
+
+    >>> array.set_annotation("sym_id", np.array([0, 0]))
+    >>> repeated, indices = repeat_box(array)
+    >>> print(repeated.sym_id)
+    [ 0  0  1  1  2  2  3  3  4  4  5  5  6  6  7  7  8  8  9  9 10 10 11 11
+     12 12 13 13 14 14 15 15 16 16 17 17 18 18 19 19 20 20 21 21 22 22 23 23
+     24 24 25 25 26 26]
+    """
+    if atoms.box is None:
+        raise BadStructureError("Structure has no box")
+
+    repeat_coord, indices = repeat_box_coord(atoms.coord, atoms.box, amount)
+    # Unroll repeated coordinates for input to 'repeat()'
+    if repeat_coord.ndim == 2:
+        repeat_coord = repeat_coord.reshape(-1, atoms.array_length(), 3)
+    else:  # ndim == 3
+        repeat_coord = repeat_coord.reshape(
+            atoms.stack_depth(), -1, atoms.array_length(), 3
+        )
+        repeat_coord = np.swapaxes(repeat_coord, 0, 1)
+
+    repeated_atoms = repeat(atoms, repeat_coord)
+    if "sym_id" in atoms.get_annotation_categories():
+        max_sym_id = np.max(atoms.sym_id)
+        # for the first repeat, (max_sym_id + 1) is added,
+        # for the second repeat 2*(max_sym_id + 1) etc.
+        repeated_atoms.sym_id += (max_sym_id + 1) * (
+            np.arange(repeated_atoms.array_length()) // atoms.array_length()
+        )
+    return repeated_atoms, indices
+
+
+def repeat_box_coord(coord, box, amount=1):
+    r"""
+    Similar to :func:`repeat_box()`, repeat the coordinates in a box by
+    duplicating and placing them in adjacent boxes.
+
+    Parameters
+    ----------
+    coord : ndarray, dtype=float, shape=(n,3) or shape=(m,n,3)
+        The coordinates to be repeated.
+    box :  ndarray, dtype=float, shape=(3,3) or shape=(m,3,3)
+        The reference box.
+        If only one box is provided, i.e. the shape is *(3,3)*,
+        the box is used for all models *m*, if the coordinates shape
+        is *(m,n,3)*.
+    amount : int, optional
+        The amount of boxes that are created in each direction of the
+        central box.
+        Hence, the total amount of boxes is
+        :math:`(1 + 2 \cdot \text{amount}) ^ 3`.
+        By default, one box is created in each direction, totalling in
+        27 boxes.
+
+    Returns
+    -------
+    repeated : ndarray, dtype=float, shape=(p,3) or shape=(m,p,3)
+        The repeated coordinates, with the same dimension as the input
+        `coord`.
+        Includes the original coordinates (central box) in the beginning
+        of the array.
+    indices : ndarray, dtype=int, shape=(p,3)
+        Indices to the coordinates in the original array.
+        Equal to
+        ``numpy.tile(np.arange(coord.shape[-2]), (1 + 2 * amount) ** 3)``.
+    """
+    if not isinstance(amount, Integral):
+        raise TypeError("The amount must be an integer")
+    # List of numpy arrays for each box repeat
+    coords_for_boxes = [coord]
+    for i in range(-amount, amount + 1):
+        for j in range(-amount, amount + 1):
+            for k in range(-amount, amount + 1):
+                # Omit the central box
+                if i != 0 or j != 0 or k != 0:
+                    temp_coord = coord.copy()
+                    # Shift coordinates to adjacent box/unit cell
+                    translation_vec = np.sum(
+                        box * np.array([i, j, k])[:, np.newaxis], axis=-2
+                    )
+                    # 'newaxis' to perform same translation on all
+                    # atoms for each model
+                    temp_coord += translation_vec[..., np.newaxis, :]
+                    coords_for_boxes.append(temp_coord)
+    return (
+        np.concatenate(coords_for_boxes, axis=-2),
+        np.tile(np.arange(coord.shape[-2]), (1 + 2 * amount) ** 3),
+    )
+
+
+def move_inside_box(coord, box):
+    r"""
+    Move all coordinates into the given box, with the box vectors
+    originating at *(0,0,0)*.
+
+    Coordinates are outside the box, when they cannot be represented by
+    a linear combination of the box vectors with scalar factors
+    :math:`0 \le a_i \le 1`.
+    In this case the affected coordinates are translated by the box
+    vectors, so that they are inside the box.
+
+    Parameters
+    ----------
+    coord : ndarray, dtype=float, shape=(n,3) or shape=(m,n,3)
+        The coordinates for one or multiple models.
+    box : ndarray, dtype=float, shape=(3,3) or shape=(m,3,3)
+        The box(es) for one or multiple models.
+        When `coord` is given for multiple models, :attr:`box` must be
+        given for multiple models as well.
+
+    Returns
+    -------
+    moved_coord : ndarray, dtype=float, shape=(n,3) or shape=(m,n,3)
+        The moved coordinates.
+        Has the same shape is the input `coord`.
+
+    Examples
+    --------
+
+    >>> box = np.array([[10,0,0], [0,10,0], [0,0,10]], dtype=float)
+    >>> inside_coord        = [ 1,  2,  3]
+    >>> outside_coord       = [ 1, 22, 54]
+    >>> other_outside_coord = [-4,  8,  6]
+    >>> coord = np.stack([inside_coord, outside_coord, other_outside_coord])
+    >>> print(coord)
+    [[ 1  2  3]
+     [ 1 22 54]
+     [-4  8  6]]
+    >>> moved_coord = move_inside_box(coord, box)
+    >>> print(moved_coord.astype(int))
+    [[1 2 3]
+     [1 2 4]
+     [6 8 6]]
+    """
+    fractions = coord_to_fraction(coord, box)
+    fractions_rem = fractions % 1
+    return fraction_to_coord(fractions_rem, box)
+
+
+def remove_pbc(atoms, selection=None):
+    """
+    Remove segmentation caused by periodic boundary conditions from each
+    molecule in the given structure.
+
+    In this process the centroid of each molecule is moved into the
+    dimensions of the box.
+    To determine the molecules the structure is required to have an
+    associated `BondList`.
+    Otherwise segmentation removal is performed on a per-chain basis.
+
+    Parameters
+    ----------
+    atoms : AtomArray, shape=(n,) or AtomArrayStack, shape=(m,n)
+        The potentially segmented structure.
+        The :attr:`box` attribute must be set in the structure.
+        An associated :attr:`bonds` attribute is recommended.
+    selection : ndarray, dtype=bool, shape=(n,)
+        Specifies which parts of `atoms` are sanitized, i.e the
+        segmentation is removed.
+
+    Returns
+    -------
+    sanitized_atoms : AtomArray or AtomArrayStack
+        The input structure with removed segmentation over periodic
+        boundaries.
+
+    See Also
+    --------
+    remove_pbc_from_coord : Variant that acts directly on coordinates.
+
+    Notes
+    -----
+    This function ensures that adjacent atoms in the input
+    :class:`AtomArray`/:class:`AtomArrayStack` are spatially close to
+    each other, i.e. their distance to each other is be smaller than the
+    half box size.
+    """
+    # Avoid circular import
+    from biotite.structure.geometry import centroid
+
+    if atoms.box is None:
+        raise BadStructureError("The 'box' attribute must be set in the structure")
+    new_atoms = atoms.copy()
+
+    if atoms.bonds is not None:
+        molecule_masks = get_molecule_masks(atoms)
+    else:
+        molecule_masks = get_chain_masks(atoms, get_chain_starts(atoms))
+
+    for mask in molecule_masks:
+        if selection is not None:
+            mask &= selection
+        # Remove segmentation in molecule
+        new_atoms.coord[..., mask, :] = remove_pbc_from_coord(
+            new_atoms.coord[..., mask, :], atoms.box
+        )
+        # Put center of molecule into box
+        center = centroid(new_atoms.coord[..., mask, :])[..., np.newaxis, :]
+        center_in_box = move_inside_box(center, new_atoms.box)
+        new_atoms.coord[..., mask, :] += center_in_box - center
+
+    return new_atoms
+
+
+def remove_pbc_from_coord(coord, box):
+    """
+    Remove segmentation caused by periodic boundary conditions from
+    given coordinates.
+
+    In this process the first coordinate is taken as origin and
+    is moved inside the box.
+    All other coordinates are assembled relative to the origin by using
+    the displacement coordinates in adjacent array positions.
+
+    Parameters
+    ----------
+    coord : ndarray, dtype=float, shape=(m,n,3) or shape=(n,3)
+        The coordinates of the potentially segmented structure.
+    box : ndarray, dtype=float, shape=(m,3,3) or shape=(3,3)
+        The simulation box or unit cell that is used as periodic
+        boundary.
+        The amount of dimensions must fit the `coord` parameter.
+
+    Returns
+    -------
+    sanitized_coord : ndarray, dtype=float, shape=(m,n,3) or shape=(n,3)
+        The reassembled coordinates.
+
+    See Also
+    --------
+    move_inside_box : The reverse operation.
+
+    Notes
+    -----
+    This function solves a common problem from MD simulation output:
+    When atoms move over the periodic boundary, they reappear on the
+    other side of the box, segmenting the structure.
+    This function reassembles the given coordinates, removing the
+    segmentation.
+    """
+
+    # Import in function to avoid circular import
+    from biotite.structure.geometry import index_displacement
+
+    # Get the PBC-sanitized displacements of all coordinates
+    # to the respective next coordinate
+    index_pairs = np.stack(
+        [np.arange(0, coord.shape[-2] - 1), np.arange(1, coord.shape[-2])], axis=1
+    )
+    neighbour_disp = index_displacement(coord, index_pairs, box=box, periodic=True)
+    # Get the PBC-sanitized displacements of all but the first
+    # coordinates to (0,0,0)
+    absolute_disp = np.cumsum(neighbour_disp, axis=-2)
+    # The first coordinate should be inside the box
+    base_coord = move_inside_box(coord[..., 0:1, :], box)
+    # The new coordinates are obtained by adding the displacements
+    # to the new origin
+    sanitized_coord = np.zeros(coord.shape, coord.dtype)
+    sanitized_coord[..., 0:1, :] = base_coord
+    sanitized_coord[..., 1:, :] = base_coord + absolute_disp
+    return sanitized_coord
+
+
+def coord_to_fraction(coord, box):
+    """
+    Transform coordinates to fractions of box vectors.
+
+    Parameters
+    ----------
+    coord : ndarray, dtype=float, shape=(n,3) or shape=(m,n,3)
+        The coordinates for one or multiple models.
+    box : ndarray, dtype=float, shape=(3,3) or shape=(m,3,3)
+        The box(es) for one or multiple models.
+        When `coord` is given for multiple models, :attr:`box` must be
+        given for multiple models as well.
+
+    Returns
+    -------
+    fraction : ndarray, dtype=float, shape=(n,3) or shape=(m,n,3)
+        The fractions of the box vectors.
+
+    See Also
+    --------
+    fraction_to_coord : The reverse operation.
+
+    Examples
+    --------
+
+    >>> box = np.array([[5,0,0], [0,5,0], [0,5,5]], dtype=float)
+    >>> coord = np.array(
+    ...     [[1,1,1], [10,0,0], [0,0,10], [-5,2,1]],
+    ...     dtype=float
+    ... )
+    >>> print(coord)
+    [[ 1.  1.  1.]
+     [10.  0.  0.]
+     [ 0.  0. 10.]
+     [-5.  2.  1.]]
+    >>> fractions = coord_to_fraction(coord, box)
+    >>> print(fractions)
+    [[ 0.2  0.0  0.2]
+     [ 2.0  0.0  0.0]
+     [ 0.0 -2.0  2.0]
+     [-1.0  0.2  0.2]]
+    """
+    return np.matmul(coord, linalg.inv(box))
+
+
+def fraction_to_coord(fraction, box):
+    """
+    Transform fractions of box vectors to coordinates.
+
+    This is the reverse operation of :func:`coord_to_fraction()`.
+
+    Parameters
+    ----------
+    fraction : ndarray, dtype=float, shape=(n,3) or shape=(m,n,3)
+        The fractions of the box vectors for one or multiple models.
+    box : ndarray, dtype=float, shape=(3,3) or shape=(m,3,3)
+        The box(es) for one or multiple models.
+        When `coord` is given for multiple models, :attr:`box` must be
+        given for multiple models as well.
+
+    Returns
+    -------
+    coord : ndarray, dtype=float, shape=(n,3) or shape=(m,n,3)
+        The coordinates.
+
+    See Also
+    --------
+    coord_to_fraction : The reverse operation.
+    """
+    return np.matmul(fraction, box)
+
+
+def is_orthogonal(box):
+    """
+    Check, whether a box or multiple boxes is/are orthogonal.
+
+    A box is orthogonal when the dot product of all box vectors with
+    each other is 0.
+
+    Parameters
+    ----------
+    box : ndarray, dtype=float, shape=(3,3) or shape=(m,3,3)
+        A single box or multiple boxes.
+
+    Returns
+    -------
+    is_orthgonal : bool or ndarray, shape=(m,), dtype=bool
+        True, if the box vectors are orthogonal, false otherwise.
+
+    Notes
+    -----
+    Due to possible numerical errors, this function also evaluates two
+    vectors as orthogonal, when their dot product is not exactly zero,
+    but it is within a small tolerance (:math:`10^{-6}`).
+    """
+    # Fix numerical errors, as values, that are actually 0,
+    # might not be calculated as such
+    tol = 1e-6
+    return (
+        (np.abs(vector_dot(box[..., 0, :], box[..., 1, :])) < tol)
+        & (np.abs(vector_dot(box[..., 0, :], box[..., 2, :])) < tol)
+        & (np.abs(vector_dot(box[..., 1, :], box[..., 2, :])) < tol)
+    )
+
+
+@functools.cache
+def _get_transformation_data():
+    with open(Path(__file__).parent / "spacegroups.json") as file:
+        return json.load(file)