1.当相乘的都为数组array时:d*f为对应元素的乘积,multiply(d,f)也是对应元素的乘积,dot(d,f)会转化为矩阵的乘积。
import numpy as np
d = np.array([[1,2],[3,4]])
f = np.array([[1,2],[3,4]])
d*f
array([[ 1, 4],
[ 9, 16]])
np.multiply(d,f)
array([[ 1, 4],
[ 9, 16]])
np.dot(d,f)
array([[ 7, 10],
[15, 22]])
2.当相乘的都为矩阵mat时:d*f为矩阵的乘积,multiply(d,f)为对应元素的乘积,dot(d,f)为矩阵的乘积。
import numpy as np
d = np.mat([[1,2],[3,4]])
f = np.mat([[1,2],[3,4]])
d*f
matrix([[ 7, 10],
[15, 22]])
np.multiply(d,f)
matrix([[ 1, 4],
[ 9, 16]])
np.dot(d,f)
matrix([[ 7, 10],
[15, 22]])
3.当相乘的既有数组array,又有矩阵mat时:混合时默认的是矩阵乘法,既d*f为矩阵的乘积,multiply(d,f)为对应元素的乘积,dot(d,f)为矩阵的乘积。
import numpy as np
d = np.array([[1,2],[3,4]])
f = np.mat([[1,2],[3,4]])
d*f
matrix([[ 7, 10],
[15, 22]])
np.multiply(d,f)
matrix([[ 1, 4],
[ 9, 16]])
np.dot(d,f)
matrix([[ 7, 10],
[15, 22]])
4.运算时最好进行矩阵的运算,因为矩阵可以方便的求转置、逆、迹。
a = np.dot(d,f)
a.T #转置
matrix([[ 7, 15],
[10, 22]])
a.I #逆
matrix([[ 5.5 , -2.5 ],
[-3.75, 1.75]])
np.trance(a) #迹
29