NumPy: numpy.broadcast_to() function

numpy.broadcast_to() function

The numpy.broadcast_to() function is used to produce an object that mimics broadcasting.

This function is useful when we want to broadcast an array to a larger shape or for arithmetic operations on arrays with different shapes.

Syntax:

numpy.broadcast_to(array, shape, subok=False)

Parameters:

Return value:

If True, then sub-classes will be passed-through, otherwise the returned array will be forced to be a base-class array (default).

Raises: ValueError - If the array is not compatible with the new shape according to NumPy’s broadcasting rules.

Example: Broadcasting an array to a larger shape using numpy.broadcast_to()

>>> import numpy as np
>>> a = np.array([2,3,4])
>>> np.broadcast_to(a, (3, 3))
array([[2, 3, 4],
       [2, 3, 4],
       [2, 3, 4]])

In the above code, an array 'a' is created with the values [2, 3, 4]. The numpy.broadcast_to() function is then used to create a new array with shape (3, 3) by repeating the values of 'a'. The resulting array has the values [[2, 3, 4], [2, 3, 4], [2, 3, 4]].

Pictorial Presentation:

Example: Broadcasting and Masked Arrays

>>> import numpy as np
>>> x = np.ma.array([2, 3, 4], mask=[False, True, False])
>>> y = np.broadcast_to(x, (3, 3), subok=True)
>>> y.mask
False
>>> y.mask = np.broadcast_to(x.mask, y.shape)
>>> y
masked_array(data =
 [[2 -- 4]
 [2 -- 4]
 [2 -- 4]],
             mask =
 [[False  True False]
 [False  True False]
 [False  True False]],
       fill_value = 999999)

In the above code, first a masked array 'x' is created using the numpy.ma.array() function. The mask parameter specifies which elements of the array are masked (in this case, the second element is masked).
Then, numpy.broadcast_to() is used to create a new array y with the same shape as a 3x3 matrix. The subok parameter is set to True to allow the output to be a masked array.
Next, the mask of y is set to a broadcasted version of the mask of x, which ensures that the masked elements are propagated to the new array.

Python - NumPy Code Editor:

Previous: broadcast
Next: broadcast_arrays()