Numpy基础8：矩阵操作

Numpy

Numpy基础8：矩阵操作

定义变量A、B

A = np.ones(shape=(5,4),dtype='int8')
B = np.full(shape=(5,4),fill_value=2,dtype='int8')

求和

np.add()

np.add(A,B)

$\Rightarrow$array([[3, 3, 3, 3], [3, 3, 3, 3], [3, 3, 3, 3], [3, 3, 3, 3], [3, 3, 3, 3]], dtype=int8)

相减

np.subtract()

np.subtract(B,A)

$\Rightarrow$array([[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]], dtype=int8)

乘积

np.multiply()

np.multiply(A,B)

$\Rightarrow$array([[2, 2, 2, 2], [2, 2, 2, 2], [2, 2, 2, 2], [2, 2, 2, 2], [2, 2, 2, 2]], dtype=int8)

相除

np.divide()

np.divide(A,B)

$\Rightarrow$array([[0.5, 0.5, 0.5, 0.5], [0.5, 0.5, 0.5, 0.5], [0.5, 0.5, 0.5, 0.5], [0.5, 0.5, 0.5, 0.5], [0.5, 0.5, 0.5, 0.5]])

点积

np.dot()

np.dot(A.T,B)

$\Rightarrow$array([[10, 10, 10, 10], [10, 10, 10, 10], [10, 10, 10, 10], [10, 10, 10, 10]], dtype=int8)

向量运算

np.add(A.T,[1,2,3,4,5])

$\Rightarrow$array([[2, 3, 4, 5, 6],[2, 3, 4, 5, 6], [2, 3, 4, 5, 6], [2, 3, 4, 5, 6]])

np.dot(A.T,[1,2,3,4,5])  #会降维

$\Rightarrow$array([15, 15, 15, 15])

---

三维张量和矩阵之间的点积

定义变量C

C = np.random.randint(0,150,(6,5,4))

当前张量中一共有6个矩阵
张量和矩阵的运算 相当于 把每一个矩阵和A.T矩阵做一个点积

np.dot(C,A.T)

$\Rightarrow$array([[[428, 428, 428, 428, 428], [336, 336, 336, 336, 336], [211, 211, 211, 211, 211],[319, 319, 319, 319, 319], [263, 263, 263, 263, 263]],…, [[188, 188, 188, 188, 188], [250, 250, 250, 250, 250], [270, 270, 270, 270, 270], [245, 245, 245, 245, 245], [435, 435, 435, 435, 435]]])

• 三维张量和向量运算(同理)