broadcast_to#

named_arrays.broadcast_to(array: float | complex, shape: dict[str, int], append: bool = False) → ScalarArray[ndarray]#
named_arrays.broadcast_to(array: NDArrayT, shape: dict[str, int], append: bool = False) → ScalarArray[NDArrayT]
named_arrays.broadcast_to(array: ArrayT, shape: dict[str, int], append: bool = False) → ArrayT

Broadcast the given array to a given shape.

Parameters:

  • array – The array to broadcast.

  • shape – The desired shape of the output array.

  • append – A boolean flag indicating whether to throw an error if there are axes in array that aren’t in shape. If append is False, the axes of array must be a subset of shape, otherwise a ValueError is raised. If append is True, the array will be broadcasted to the shape: na.broadcast_shapes(array.shape, shape).

See also

numpy.broadcast_to()

Equivalent Numpy function.

Examples

If we define some array shapes as

import named_arrays as na

shape_x = dict(x=3)
shape_y = dict(y=4)

and a random 1D array as an example

a = na.random.uniform(0, 1, shape_x)

Then we can broadcast it to 2 dimensions using the union of shape_x and shape_y

na.broadcast_to(a, shape_x | shape_y)

ScalarArray(
    ndarray=[[0.93703776, 0.93703776, 0.93703776, 0.93703776],
             [0.98034735, 0.98034735, 0.98034735, 0.98034735],
             [0.49277887, 0.49277887, 0.49277887, 0.49277887]],
    axes=('x', 'y'),
)

Alternatively, we can set the append keyword to True so that we only need to provide shape_y since the shape of the array is already shape_x.

na.broadcast_to(a, shape_y, append=True)

ScalarArray(
    ndarray=[[0.93703776, 0.93703776, 0.93703776, 0.93703776],
             [0.98034735, 0.98034735, 0.98034735, 0.98034735],
             [0.49277887, 0.49277887, 0.49277887, 0.49277887]],
    axes=('x', 'y'),
)