Python---BroadCast

一. Numpy

1. Numpy创建 2*2*3的数据

code:

import numpy as np
l1 = np.arange(12)
l1 = np.reshape(l1, (2,2,3))
print l1

output:

[[[ 0  1  2]
  [ 3  4  5]]

 [[ 6  7  8]
  [ 9 10 11]]]

2.多维数据每一位加1

code:

import numpy as np
l1 = np.arange(12)
l1 = np.reshape(l1, (2,2,3))
print l1 + 1

output

[[[ 1  2  3]
  [ 4  5  6]]

 [[ 7  8  9]
  [10 11 12]]]

3.数据在最后一维加 [1,2,3]

code:

import numpy as np
l1 = np.arange(12)
l1 = np.reshape(l1, (2,2,3))
print l1 + [1,2,3]

output:

[[[ 1  3  5]
  [ 4  6  8]]

 [[ 7  9 11]
  [10 12 14]]]

二. Tensorflow

注意:例如tf.add()接口同+号一样具有boradcast
1.创建 2*2*3的数据

import tensorflow as tf
import numpy as np

l1 = np.arange(12)
l1 = np.reshape(l1, (2,2,3))
c1 = tf.constant(l1)

with tf.Session() as sess:
    r1 = sess.run(c1)
    print r1

[[[ 0  1  2]
  [ 3  4  5]]

 [[ 6  7  8]
  [ 9 10 11]]]

2.数据每一位加1

import tensorflow as tf
import numpy as np

l1 = np.arange(12)
l1 = np.reshape(l1, (2,2,3))
c1 = tf.constant(l1)
c1 = c1 + 1

with tf.Session() as sess:
    r1 = sess.run(c1)
    print r1

[[[ 1  2  3]
  [ 4  5  6]]

 [[ 7  8  9]
  [10 11 12]]]

3.数据在最后一维加 [1,2,3]

import tensorflow as tf
import numpy as np

l1 = np.arange(12)
l1 = np.reshape(l1, (2,2,3))
c1 = tf.constant(l1)
c1 = c1 + [1,2,3]

with tf.Session() as sess:
    r1 = sess.run(c1)
    print r1

[[[ 1  3  5]
  [ 4  6  8]]

 [[ 7  9 11]
  [10 12 14]]]