如何在Numpy中方便地将range（10）分配给形状为（10，1）的ndarray？

Question

I tried this:

import numpy as np
a = np.empty((1, 10, 1), np.int8)
a[0] = range(10)

It throw error: ValueError: could not broadcast input array from shape (10) into shape (10,1)

Answer 1

您可以执行a[0, :, 0] = range(10) 。

Answer 2

Several options that work in this case:

a[0, :, 0] = np.arange(10)  # assign to 1D slice

a[0].flat = range(10)  # assign to flattened 2D slice

a[0] = np.arange(10).reshape(10, 1)  # bring the rigth side into correct shape

a[0] = np.arange(10)[:, np.newaxis]  # bring the rigth side into correct shape

Note the use of np.arange instead of range . The former directly creates an ndarray with a sequence of values, while the latter creates an iterable that needs to be converted into an array for the assignment.

In the case of assigning to flat it makes sense to use range because both are iterators.