如何在 "for loop" 中插入不同的列表切片？

Question

我正在尝试为大小为 9x9 的 numpy 数组的每一行、每一列和每一字段创建一个字典。（就像在数独游戏中一样）能够查看是否没有重复项。一切正常，但我想知道是否可以在循环中创建 field_dic？:

sudoku =[]  # Some 9x9 numpy.ndarray

row_dic = {}
col_dic = {}
field_dic = {}


for i in range(9):
    col_dic[f"Row{i}"] = dict(Counter(sudoku[:, i]))
    row_dic[f"Col{i}"] = dict(Counter(sudoku[i, :]))

field_dic = {
    "Field0": dict(Counter(list(sudoku[:3, :3].flat))),
    "Field1": dict(Counter(list(sudoku[:3, 3:6].flat))),
    "Field2": dict(Counter(list(sudoku[:3, 6:].flat))),
    "Field3": dict(Counter(list(sudoku[3:6, :3].flat))),
    "Field4": dict(Counter(list(sudoku[3:6, 3:6].flat))),
    "Field5": dict(Counter(list(sudoku[3:6, 6:].flat))),
    "Field6": dict(Counter(list(sudoku[6:, :3].flat))),
    "Field7": dict(Counter(list(sudoku[6:, 3:6].flat))),
    "Field8": dict(Counter(list(sudoku[6:, 6:].flat)))}

Answer 1

您可以使用循环，但听写理解可能是更好的选择。

让您抢先一步：

import itertools

ind_groups = [(0, 3), (3, 6), (6, 9)]
field_positions = itertools.product(ind_groups, ind_groups)
field_dict = {
  f'Field{i}': (r0, r1, c0, c1) for (i, ((r0, r1), (c0, c1))) in enumerate(field_positions)
}

Answer 2

您运行将原始数组从 (9,9) 重塑为 (3,3,3,3)。然后，重塑数组的维度 0 将为您提供行，维度 2 将为您提供块的列位置。

import numpy as np

sample = np.arange(81).reshape((9,9))
reshaped = sample.reshape(3,3,3,3)

field_dic = {}
for i in range(9):
    field_dic[f"Field{i}"] = reshaped[(i//3),:,i%3,:].reshape(9)

结果如下：

>>> print('first block:\n', reshaped[0,:,0,:])
 first block:
 [[ 0  1  2]
  [ 9 10 11]
  [18 19 20]]

>>> print(field_dic['Field0'])
[ 0  1  2  9 10 11 18 19 20]

>>> print(field_dic['Field1'])
[ 3  4  5 12 13 14 21 22 23]

>>> print(field_dic['Field3'])
[27 28 29 36 37 38 45 46 47]

如何在 "for loop" 中插入不同的列表切片？

How do I insert different list slice in "for loop"?

python

for-loop

numpy

slice