机器学习笔记--1.1 Numpy矩阵运算

1.矩阵的初始化

import numpy as np
#(1)创建一个3*5的全0和全1矩阵
myZero = np.zeros([3, 5])
myOnes = np.ones([3, 5])
#(2)生成随机矩阵
myRand = np.random.rand(3, 4)
#(3)单位阵
myEye = np.eye(3) #3*3的单位阵

2.矩阵的元素运算

from numpy import *
#(1)元素相加和相减
myOnes = ones([3, 3])
myEye = eye(3)
print (myOnes + myEye)
print (myOnes - myEye)
#(2)矩阵数乘：一个数乘以一个矩阵
mymatrix = mat([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
a = 10
print(a * mymatrix)
#(3)矩阵所有元素求和
mymatrix = mat([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
#(4)矩阵各元素的积：矩阵的点乘同维对应元素的相乘。当矩阵的维度不同时，会根据一定的广播规则将维数 
#扩充到一致的形式。
mymatrix1 = mat([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
mymatrix2 = 1.5 * ones([3, 3])
print(multiply(mymatrix1, mymatrix2))
#(5)矩阵各元素的n次幂：n=5
mylist = mat([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
print(power(mylist, 2))

3.矩阵的乘法：矩阵乘矩阵

import numpy as np
tensor1 = np.arrary([[1, 2, 3],[4, 5, 6],[7, 8, 9]])
tensor2 = np.arrary([[1, 2, 3], [1, 2, 3] [1, 2, 3]])
dot_product = tensor1.dot(tensor2)
print(dot_product)

        输出结果

        [[14, 14, 14],

        [32, 32, 32],

        [50, 50, 50]]

4.矩阵的转置

from numpy import *
mymatrix = mat([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
print mymatrix.T       #转置
mymatrix.transpose()   #转置

5.矩阵的其他操作：行列数、切片、复制、比较

from numpy import *
mymatrix = mat([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
[m, n] = shape(mymatrix)  #矩阵的行列数
print("矩阵的行数和列数：", m, n)

myscl1 = mymatrix[0]    #按行切片
print("按行切片：", myscl1)

myscl2 = mymatrix.T[0]  #按列切片
print("按列切片：", myscl2)

mycpmat = mymatrix.copy()
print("复制矩阵：", mycpmat)

#比较
print("矩阵元素的比较：\n", mymatrix < mymatrix.T)

        输出结果：

        矩阵的行数和列数： 3 3
        按行切片： [[1 2 3]]
        按列切片： [[1 4 7]]
        复制矩阵： [[1 2 3]
                            [4 5 6]
                            [7 8 9]]
        矩阵元素的比较：
         [[False  True  True]
         [False False  True]
         [False False False]]