详解Python list 与 NumPy.ndarry 切片之间的对比

# list 切片返回的是不原数据，对新数据的修改不会影响原数据
In [45]: list1 = [1, 2, 3, 4, 5] 

In [46]: list2 = list1[:3]

In [47]: list2
Out[47]: [1, 2, 3]

In [49]: list2[1] = 1999

# 原数据没变
In [50]: list1
Out[50]: [1, 2, 3, 4, 5]

In [51]: list2
Out[51]: [1, 1999, 3]



# 而 NumPy.ndarry 的切片返回的是原数据
In [52]: arr = np.array([1, 2, 3, 4, 5])

In [53]: arr
Out[53]: array([1, 2, 3, 4, 5])

In [54]: arr1 = arr[:3]

In [55]: arr1
Out[55]: array([1, 2, 3])

In [56]: arr1[0] = 989

In [57]: arr1
Out[57]: array([989,  2,  3])

# 修改了原数据
In [58]: arr
Out[58]: array([989,  2,  3,  4,  5])

# 若希望得到原数据的副本， 可以用 copy()
In [59]: arr2 = arr[:3].copy()

In [60]: arr2
Out[60]: array([989,  2,  3])

In [61]: arr2[1] = 99282

In [62]: arr2
Out[62]: array([ 989, 99282,   3])

# 原数据没被修改
In [63]: arr
Out[63]: array([989,  2,  3,  4,  5])