Numpy使我们能够高效地工作在Python中的阵列(arrays)和矩阵(matrices)。
下面是最基础的Numpy知识，这个帖子将会进行长期补充。

Start A Array

可以建立一个列表List, 并将其转换为阵列Array

np.array(mylist)
mylist = [1, 2, 3]
x = np.array(mylist)
x
>>>array([1, 2, 3])

或者更直接

y = np.array([4, 5, 6])
m = np.array([[7, 8, 9], [10, 11, 12]])

对于arange函数，我们传递一个开始start，一个停止stop和一个跨步step的值， 并在给定的间隔内返回均匀跨步的值。

n = np.arange(0, 30, 2) # start at 0 count up by 2, stop before 30
>>>
array([ 0,  2,  4,  6,  8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28])

想把这个array 转换成一个3x5的阵列

n = n.reshape(3, 5) # reshape array to be 3x5
array([[ 0,  2,  4,  6,  8],
       [10, 12, 14, 16, 18],
       [20, 22, 24, 26, 28]])

linspace函数类似于arange，除了告诉需要返回多少个数字，它会相应地分隔间隔。

o = np.linspace(0, 4, 9) # return 9 evenly spaced values from 0 to 4
o
>>> array([ 0. ,  0.5,  1. ,  1.5,  2. ,  2.5,  3. ,  3.5,  4. ])

Numpy还有几个内置函数和快捷方式来创建阵列。

• ones返回一个都是1的阵列
• zeros是一个都是0的阵列
• eye返回一个阵列，其中对角线是1，其他是0的
• diag构造一个对角阵列

np.ones((3, 2))
>>>
array([[ 1.,  1.],
       [ 1.,  1.],
       [ 1.,  1.]])

np.zeros((2, 3))
>>>
array([[ 0.,  0.,  0.],
       [ 0.,  0.,  0.]])


np.eye(3)
>>>
array([[ 1.,  0.,  0.],
       [ 0.,  1.,  0.],
       [ 0.,  0.,  1.]])

y = np.array([4, 5, 6])
np.diag(y)
>>>
array([[4, 0, 0],
       [0, 5, 0],
       [0, 0, 6]])

索引(index)和切片(slice)

创建一个array，0到12的每个数都进行平方

s = np.arange(13)**2
s
>>> array([  0,   1,   4,   9,  16,  25,  36,  49,  64,  81, 100, 121, 144])

• 用方括号，里面是数字 获取特定索引的值

s[0], s[4], s[-1]
>>>(0, 16, 144)

• 用冒号(:)符号获取范围
• 第一个例子从索引1开始的范围， 并在索引5之前停止

s[1:5]
>>>array([ 1,  4,  9, 16])

• 用负数从array的末尾倒数
• 指定起始或结束索引不是必需的，直接 ： 就不写了
• array 的最后四个元素

s[-4:]
>>>array([ 81, 100, 121, 144])

• array 的末尾第五开始， 每步向后倒退两位
  A second : can be used to indicate step-size. array[start:stop:stepsize]

s[-5::-2]
>>> array([64, 36, 16,  4,  0])

Multidimensional array.

r = np.arange(36)
r.resize((6, 6))
r
>>>
array([[ 0,  1,  2,  3,  4,  5],
       [ 6,  7,  8,  9, 10, 11],
       [12, 13, 14, 15, 16, 17],
       [18, 19, 20, 21, 22, 23],
       [24, 25, 26, 27, 28, 29],
       [30, 31, 32, 33, 34, 35]])

#Use bracket notation to slice: array[row, column]
r[2, 2]
>>> 14

# select a range of rows or columns
r[3, 3:6]
>>>
array([21, 22, 23])

• 直到前两列，直到最后一行

#selecting all the rows up to (and not including) row 2
#and all the columns up to (and not including) the last column.
r[:2, :-1]
>>>
array([[ 0,  1,  2,  3,  4],
       [ 6,  7,  8,  9, 10]])

r[-1, ::2]
>>>
array([30, 32, 34])

• 方括号运算符进行条件索引

# conditional indexing

r[r > 30]
>>>
array([31, 32, 33, 34, 35])

r[r > 30] = 30
r
>>>
array([[ 0,  1,  2,  3,  4,  5],
       [ 6,  7,  8,  9, 10, 11],
       [12, 13, 14, 15, 16, 17],
       [18, 19, 20, 21, 22, 23],
       [24, 25, 26, 27, 28, 29],
       [30, 30, 30, 30, 30, 30]])

Copy Data

• 创建一个新的new array r2，它是array r的一个切片
• 把这个阵列的所有元素设置为零
• 看原始阵列r时， 可以看到r中的slice也被改变了
• 所以这是需要记住的， 在使用Numpy阵列时要小心: 如果我们希望创建副本，但不更改原始阵列r， 可以使用r.copy()

r2 = r[:3,:3]
r2
>>>
array([[ 0,  1,  2],
       [ 6,  7,  8],
       [12, 13, 14]])

r2[:] = 0
r2
>>>
array([[0, 0, 0],
       [0, 0, 0],
       [0, 0, 0]])

r
>>>
array([[ 0,  0,  0,  3,  4,  5],
       [ 0,  0,  0,  9, 10, 11],
       [ 0,  0,  0, 15, 16, 17],
       [18, 19, 20, 21, 22, 23],
       [24, 25, 26, 27, 28, 29],
       [30, 30, 30, 30, 30, 30]])

# use r.copy to create a copy that will not affect the original array
r_copy = r.copy()
r_copy
>>>
array([[ 0,  0,  0,  3,  4,  5],
       [ 0,  0,  0,  9, 10, 11],
       [ 0,  0,  0, 15, 16, 17],
       [18, 19, 20, 21, 22, 23],
       [24, 25, 26, 27, 28, 29],
       [30, 30, 30, 30, 30, 30]])


r_copy[:] = 10
print(r_copy, '\n')
print(r)
>>>
[[10 10 10 10 10 10]
 [10 10 10 10 10 10]
 [10 10 10 10 10 10]
 [10 10 10 10 10 10]
 [10 10 10 10 10 10]
 [10 10 10 10 10 10]] 

[[ 0  0  0  3  4  5]
 [ 0  0  0  9 10 11]
 [ 0  0  0 15 16 17]
 [18 19 20 21 22 23]
 [24 25 26 27 28 29]
 [30 30 30 30 30 30]]

• 将r_copy中所有元素的值更改为10， 则原始阵列r保持不变

Iterating Over Arrays

test = np.random.randint(0, 10, (4,3))
test
>>>
array([[4, 8, 7],
       [7, 8, 3],
       [9, 1, 3],
       [7, 5, 4]])

#Iterate by row:
for row in test:
    print(row)
>>>
[4 8 7]
[7 8 3]
[9 1 3]
[7 5 4]

#Iterate by index:
for i in range(len(test)):
    print(test[i])
>>>
[4 8 7]
[7 8 3]
[9 1 3]
[7 5 4]

#Iterate by row and index:
for i, row in enumerate(test):
    print('row', i, 'is', row)
>>>
row 0 is [4 8 7]
row 1 is [7 8 3]
row 2 is [9 1 3]
row 3 is [7 5 4]

#Use zip to iterate over multiple iterables
test2 = test**2
test2
>>>
array([[16, 64, 49],
       [49, 64,  9],
       [81,  1,  9],
       [49, 25, 16]])

for i, j in zip(test, test2):
    print(i,'+',j,'=',i+j)
>>>
[4 8 7] + [16 64 49] = [20 72 56]
[7 8 3] + [49 64  9] = [56 72 12]
[9 1 3] + [81  1  9] = [90  2 12]
[7 5 4] + [49 25 16] = [56 30 20]