理解numpy中的切片访问

一维数组的切片访问

numpy 中的一维数组的切片方法与 python 内置的list 切片类似.

Details：

    1. ndarray[start:stop:step]     # means start from start, stop at stop, step by step
    2. ndarray[start:stop]          # means start from start, stop at stop, step by 1
    3. ndarray[start:]              # means start from start, stop at the end, step by 1
    4. ndarray[:stop]               # means start from the beginning, stop at stop, step by 1
    5. ndarray[:]                   # means start from the beginning, stop at the end, step by 1                   
    6. ndarray[start:stop:step, start:stop:step]                  # means 2-dimensional slicing

    Note:
    start is from 0
    start is inclusive, stop is exclusive
    step could be negative, which means reverse order

下面是例子
先构建1个1维数组：

    arr = np.arange(10) # [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
    logger.info(f"arr: {arr}") # [0 1 2 3 4 5 6 7 8 9]

1. 如果stop index > 数组长度， 则只会返回 start: last index + 1

logger.info(f"arr[1:100]: {arr[1:100]}") # [1 2 3 4 5 6 7 8 9] even stop > len(arr)

2. 正常[1:5] 则返回 index 为 1 到 4的元素， 就是 第2 到 第5 的元素

logger.info(f"arr[1:5]: {arr[1:5]}") # [1 2 3 4]

3. [1:5:2] 则返回 index 为 1 到 4的元素内， 每2个选1个，因为step = 2, 所以就是第2 到第4的元素了

logger.info(f"arr[1:5:2]: {arr[1:5:2]}") # [1 3]

4. [1:] 没有指定stop 和 step, 则stop = last index + 1 (9 + 1) , step = 1 返回第2 个到 第10个元素 (index from 1 to 9)

logger.info(f"arr[1:]: {arr[1:]}")  # [1 2 3 4 5 6 7 8 9]

5. [:5] 没有指定start 和 step, 则stop = 0 , step = 1 返回第1 到 第5 个元素 (index from 0 to 4)

logger.info(f"arr[:5]: {arr[:5]}") # [0 1 2 3 4]

6. [:] 返回所有元素， 等于[0: len(arr):1]

logger.info(f"arr[:]: {arr[:]}") #  [0 1 2 3 4 5 6 7 8 9]

7. [::-1] 返回所有元素, 反向倒置， 因为step = -1

logger.info(f"arr[::-1]: {arr[::-1]}") # [9 8 7 6 5 4 3 2 1 0]

8. [5:2:-1] 返回 从 index 5 到 index 3(不包含2) 中, 反向相反， 每隔1个的元素

logger.info(f"arr[5:1:-2]: {arr[5:1:-2]}") # [5 3]

8. [5:2] 返回空数组， 因为start > stop 且step > 0

logger.info(f"arr[5:1]: {arr[5:1]}") # []  # empty array because start > stop

二维/多维数组的切片访问

规则， 其实是基于1维数组的

arr[0轴上的切片， 1轴上的切片…， n-1轴上的切片]

例子：

    # 2-dimensional slicing
    """
    [[ 0  1  2  3]
     [ 4  5  6  7]
     [ 8  9 10 11]]
    """
    arr = np.arange(12).reshape(3, 4)
    logger.info(f"arr: {arr}")
    logger.info(f"arr[1:3, 1:3]: {arr[1:3, 1:3]}") # means row 1, 2, column 1, 2 # [[ 5  6] [ 9 10]]
    logger.info(f"arr[1:3, :]: {arr[1:3, :]}") # means row 1, 2, all columns # [[ 4  5  6  7] [ 8  9 10 11]]
    logger.info(f"arr[:, 1:3]: {arr[:, 1:3]}") # means all rows, column 1, 2 # [[ 1  2] [ 5  6] [ 9 10]]