NumPy.array 求点乘， 向量长度， 向量夹角的方法

求点乘方法有：

1. np.sum(a*b)

2. (a*b).sum()

3. np.dot(a, b)

4. a.dot(b)

5. b.dot(a)

求向量长度方法有：

1. np.sqrt((a*a).sum()) 

2. np.linalg.norm(a)

求两个向量夹角：

根据点乘定义，cosangle = a.dot(b)/(np.linalg.norm(a) * np.linalg.norm(b))  

In [25]: a = np.array([1,2])                                                    

In [26]: b = np.array([2, 1])                                                   

In [27]: dot = 0                                                                

In [28]: for e, f in zip(a, b): 
    ...:     dot += e*f 
    ...:                                                                        

In [29]: dot                                                                    
Out[29]: 4

In [30]: a*b                                                                    
Out[30]: array([2, 2])

In [31]: np.sum(a*b)                                                            
Out[31]: 4

In [32]: (a*b).sum()                                                            
Out[32]: 4

In [33]: np.dot(a,b)                                                            
Out[33]: 4

In [34]: a.dot(b)                                                               
Out[34]: 4

In [35]: b.dot(a)                                                               
Out[35]: 4

In [36]: amag = np.sqrt((a*a).sum())                                            

In [37]: amag                                                                   
Out[37]: 2.23606797749979

In [38]: amag = np.linalg.norm(a)                                               

In [39]: amag                                                                   
Out[39]: 2.23606797749979

In [40]: cosangle = a.dot(b)/(np.linalg.norm(a) * np.linalg.norm(b))            

In [41]: cosangle                                                               
Out[41]: 0.7999999999999998

In [42]: angle = np.arccos(cosangle)                                            

In [43]: angle                                                                  
Out[43]: 0.6435011087932847