如何递归放大 Numpy 数组

Question

我试图证明我生成的数组在多个长度尺度上看起来是自相似的。为此，我想编写一个函数来打印输入数组的左上角，然后打印该部分的左上角，依此类推，以指定迭代次数。我写了这个递归函数。但是，输出不是我所期望的。

def zoom(array, max_zoom, zoom_level):
    half_width = int(array.shape[0]/2)
    half_height = int(array.shape[1]/2)
    print(array)

    while zoom_level < max_zoom:
        array = array[:half_width, :half_height]
        zoom_level += 1
        zoom(array, max_zoom, zoom_level)

问题不是打印到所需的级别然后停止，而是数组开始再次变大，并且在几次额外的迭代后函数停止。

例如 zoom(array, 3, 0) 应该在下面给出 8x8 网格的输出。

[[0. 0. 0. 0. 0. 0. 0. 0.]
[0. 0. 0. 0. 0. 0. 0. 0.]
[0. 0. 0. 0. 0. 0. 0. 0.]
[0. 0. 0. 0. 0. 0. 0. 0.]
[0. 0. 0. 0. 0. 0. 0. 0.]
[0. 0. 0. 0. 0. 0. 0. 0.]
[0. 0. 0. 0. 0. 0. 0. 0.]
[0. 0. 0. 0. 0. 0. 0. 0.]] 

[[0. 0. 0. 0.]
[0. 0. 0. 0.]
[0. 0. 0. 0.]
[0. 0. 0. 0.]]

[[0. 0.]
[0. 0.]]

[[0.]]

而是给出：

[[0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0.]]
[[0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]]
[[0. 0.]
 [0. 0.]]
[[0.]]
[[0. 0.]
 [0. 0.]]
[[0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]]
[[0. 0.]
 [0. 0.]]
[[0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]]

任何人都可以阐明为什么此功能行为不端吗？

Answer 1

问题在于您的递归函数有一段时间。您正在做的是为每个递归调用重复相同的过程，从而打印出不必要的结果。

def zoom(array, max_zoom, zoom_level):
    print(array)
    if zoom_level < max_zoom:
        half_width = array.shape[0] // 2
        half_height = array.shape[1] // 2
        array = array[:half_width, :half_height]
        zoom_level += 1
        zoom(array, max_zoom, zoom_level)

A = np.zeros((9, 9))

zoom(A, 3, 0)

输出将是：

[[0. 0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 0. 0.]]
[[0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]]
[[0. 0.]
 [0. 0.]]
[[0.]]