Array Basics

import numpy as np

Create Arrays

The basic array type in NumPy is ndarray. It is also known by the alias array. So when we talk about array, we are referring to the ndarray unless specified.

An ndarray object stores a matrix, every elements of which have the same data type. The matrix is indexed by a tuple of non-negative integers.

Dimensions of matrices are called axes (axis).

np.array(
    object, 
    dtype = None, 
    copy = True, 
    order = None, 
    subok = False, 
    ndmin = 0
)
"""
#pointer, dtype, shape, stride
#object: 
    array; 
    object exposing the array interface;
    object whose __array__ returns an array
    nested list
#dtype: data type
#copy: if this is a copy
"""

a = np.array([[1,2],[3,4]])
b = np.array([1+1.j,2+1.j],dtype = np.complex)  # a complex array

# use function of coordinates to create an array from a given shape
np.fromfunction(lambda i, j: i == j, (3, 3), dtype=int)
array([[ True, False, False],
       [False,  True, False],
       [False, False,  True]])

Properties

>>> a = np.array([[1,2],[3,4],[5,6]])
>>> a.ndim
2
>>> a.shape
(3, 2)
>>> a.size
6
>>> a.dtype
dtype('int32')
>>> a.dtype.name
'int32'
>>> print(a)
[[1 2]
 [3 4]
 [5 6]]

• data types for np:

  • bool

  • int8 int16 int32 int64

  • uint8 uint16 uint32 uint64

  • float float16 float32 float64

  • complex64 complex128

dt = np.dtype(np.int32) # this is an np.dtype object

Special Arrays

Notice that following methods all return array objects.

np.empty(shape) # 未初始化
np.zeros(shape) # 全零数组
np.ones(shape) # 全1数组
np.asarray(data) # 从已有的矩阵结构数据创建数组

# create sequences:
# [x for x in range(start,stop) if x-start % step == 0]:
np.arrange(start,stop,step,dtype)
# [start+x*(stop-start)/(num-1) for x in range(num)]
np.linspace(start,stop,num,endpoint=True)
# [base**(start+x*(stop-start)/(num-1)) for x in range(num)]
np.logspace(start,stop,num,endpoint=True,base=10.0)

# create a 1-dim array from an object exposing the buffer interface
np.frombuffer(buffer)

# create a 1-dim array from an iterable object
np.fromiter(iterable)

Indexing, Slicing & Iterating

a = np.array([[1,2],[3,4],[5,6]])

# Indexing:
a[2, 3]

# Slicing
a[0:2, 1:2] # [[2],[4]]
a[0:2, 1] # [2, 4]
# use ... or : to take the whole slice
a[:, 1] # [2,4,6]

# 取a的[0,0],[1,1],[2,0]索引并返回数列
a[[0,1,2],[0,1,0]] # [1,4,5]

#数组Iteration:
#nditer迭代
for x in np.nditer(a):
    #...
#默认按内存存储顺序访问，指定C风格为行优先，Fortran风格为列优先
order = 'C'
order = 'F'
#通过指定读取模式修改数组元素：
op_flags=['readwrite']
#flat迭代
for x in a.flat:
    #...

Arithmetic Operations

b = a.reshape(2,3)
b = np.transpose(a,axes)#对换
b = a.T() #转置
b = np.swapaxes(a,axis1,axis2)#对应轴的索引，从0开始算
b = np.expand_dims(a,axis)#扩展一个新轴
b = np.squeeze(a,axis)#删除长度为1的维度
b = np.concatenate((arr1,arr2,...),axis)#沿axis连接相同宽度的arr形成新数组
b = np.stack((arr1,arr2,...),axis)#沿axis堆叠相同形状的arr形成新数组

np.add(a,b) or a+b
np.subtract(a,b) or a-b
np.multiply(a,b) or a*b
np.divide(a,b) or a/b 
np.power(a,b) or a**b
np.mod(a,b) or a%b
#以上都是逐项运算，要求同型
np.dot(a,b) or a@b #矩阵乘积
np.linalg.det(a)
np.linalg.solve(a)
np.linalg.inv(a)

np.sum(a, axis=None, dtype=None, out=None, keepdims=np._NoValue)

See also Python - Emulating numeric types.

Reference

[1] https://numpy.org/devdocs/user/quickstart.html