mxnet中ndarra基础用法

from mxnet import nd

x = nd.arange(12)
x

[ 0.  1.  2.  3.  4.  5.  6.  7.  8.  9. 10. 11.]
<NDArray 12 @cpu(0)>

x.shape

(12,)

x.size

12

x = x.reshape(3,4)
x

[[ 0.  1.  2.  3.]
 [ 4.  5.  6.  7.]
 [ 8.  9. 10. 11.]]
<NDArray 3x4 @cpu(0)>

nd.zeros((2,3))

[[0. 0. 0.]
 [0. 0. 0.]]
<NDArray 2x3 @cpu(0)>

nd.ones((2,3,4))

[[[1. 1. 1. 1.]
  [1. 1. 1. 1.]
  [1. 1. 1. 1.]]

 [[1. 1. 1. 1.]
  [1. 1. 1. 1.]
  [1. 1. 1. 1.]]]
<NDArray 2x3x4 @cpu(0)>

y = nd.array([[1,2,3,4],[5,6,7,8],[4,3,2,1]])
y

[[1. 2. 3. 4.]
 [5. 6. 7. 8.]
 [4. 3. 2. 1.]]
<NDArray 3x4 @cpu(0)>

#随机生成均值为0,标准差为1的正态分布数据
nd.random.normal(0,1,shape=(3,4))

[[ 2.2122064   0.7740038   1.0434403   1.1839255 ]
 [ 1.8917114  -1.2347414  -1.771029   -0.45138445]
 [ 0.57938355 -1.856082   -1.9768796  -0.20801921]]
<NDArray 3x4 @cpu(0)>

#以下是ndarray的运算

x + y

[[ 1.  3.  5.  7.]
 [ 9. 11. 13. 15.]
 [12. 12. 12. 12.]]
<NDArray 3x4 @cpu(0)>

x*y

[[ 0.  2.  6. 12.]
 [20. 30. 42. 56.]
 [32. 27. 20. 11.]]
<NDArray 3x4 @cpu(0)>

x/y

[[ 0.          0.5         0.6666667   0.75      ]
 [ 0.8         0.8333333   0.85714287  0.875     ]
 [ 2.          3.          5.         11.        ]]
<NDArray 3x4 @cpu(0)>

y.exp()

[[2.7182817e+00 7.3890562e+00 2.0085537e+01 5.4598148e+01]
 [1.4841316e+02 4.0342880e+02 1.0966332e+03 2.9809580e+03]
 [5.4598148e+01 2.0085537e+01 7.3890562e+00 2.7182817e+00]]
<NDArray 3x4 @cpu(0)>

#x:3*4 , y.T:4*3 , 结果：3×3
nd.dot(x,y.T)

[[ 20.  44.  10.]
 [ 60. 148.  50.]
 [100. 252.  90.]]
<NDArray 3x3 @cpu(0)>

#连接ndarray,分别是在维度0和维度1上连接
nd.concat(x , y,dim=0),nd.concat(x,y,dim=1)

(
 [[ 0.  1.  2.  3.]
  [ 4.  5.  6.  7.]
  [ 8.  9. 10. 11.]
  [ 1.  2.  3.  4.]
  [ 5.  6.  7.  8.]
  [ 4.  3.  2.  1.]]
 <NDArray 6x4 @cpu(0)>,
 
 [[ 0.  1.  2.  3.  1.  2.  3.  4.]
  [ 4.  5.  6.  7.  5.  6.  7.  8.]
  [ 8.  9. 10. 11.  4.  3.  2.  1.]]
 <NDArray 3x8 @cpu(0)>)

#判别式，对应值相等返回1,否则0
x == y

[[0. 0. 0. 0.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]]
<NDArray 3x4 @cpu(0)>

x.sum()

[66.]
<NDArray 1 @cpu(0)>

#求范数
x.norm()

[22.494442]
<NDArray 1 @cpu(0)>

#将ndarray转换为标量
x.sum().asscalar()

66.0

#广播机制
a = nd.arange(3).reshape(3,1)
b = nd.arange(2).reshape(1,2)
a,b

(
 [[0.]
  [1.]
  [2.]]
 <NDArray 3x1 @cpu(0)>,
 
 [[0. 1.]]
 <NDArray 1x2 @cpu(0)>)

a+b

[[0. 1.]
 [1. 2.]
 [2. 3.]]
<NDArray 3x2 @cpu(0)>

#索引
x[1:3]

[[ 4.  5.  6.  7.]
 [ 8.  9. 10. 11.]]
<NDArray 2x4 @cpu(0)>

x[1,2]=9
x

[[ 0.  1.  2.  3.]
 [ 4.  5.  9.  7.]
 [ 8.  9. 10. 11.]]
<NDArray 3x4 @cpu(0)>

x[1:2,:]=12
x

[[ 0.  1.  2.  3.]
 [12. 12. 12. 12.]
 [ 8.  9. 10. 11.]]
<NDArray 3x4 @cpu(0)>

#ndarray和numpy的转换
import numpy as np

p = np.ones((1,2))
d = nd.array(p)
d

[[1. 1.]]
<NDArray 1x2 @cpu(0)>

d.asnumpy()

array([[1., 1.]], dtype=float32)