np.diff()的用法

numpy.diff(a, n=1,axis=-1)

沿着指定軸計算第N維的離散差值
參數：
a：輸入矩陣
n：可選，代表要執行幾次差值
axis：默認是最後一個

In [21]:a = np.random.randint(0,9,(3,4))

In [22]: a
Out[22]: 
array([[6, 4, 6, 5],
       [0, 4, 6, 1],
       [7, 2, 5, 0]])

In [23]: a[0,:]
Out[23]: array([6, 4, 6, 5])

In [24]: a[-1,:]
Out[24]: array([7, 2, 5, 0])

In [25]: m = a[0,:]-a[-1,:]

In [26]: n = np.diff(a,n=1,axis=0)

In [27]: m
Out[27]: array([-1,  2,  1,  5])

In [28]: n
Out[28]: 
array([[-6,  0,  0, -4],
       [ 7, -2, -1, -1]])

In [30]: v = np.vstack((n,m))

In [31]: v
Out[31]: 
array([[-6,  0,  0, -4],
       [ 7, -2, -1, -1],
       [-1,  2,  1,  5]])


In [32]: n1 = np.diff(a,n=2,axis=0)

In [34]: n
Out[34]: 
array([[-6,  0,  0, -4],
       [ 7, -2, -1, -1]])

In [35]: n1
Out[35]: array([[13, -2, -1,  3]])