python踩坑记1矩阵乘法

python踩坑记1矩阵乘法@TOC

本人自学的python,从照书抄代码，到看视频，上社区，踩了不少坑，其中有版本原因、有书写不规范，在此记录下来，也与各位新人共免，希望大虾们不要取笑才好。
1、tensorflow 张量运算问题
TensorFlow快速入门教程中矩阵相加问题

t_2 = tf.constant([4,3,2])
t_1 = tf.constant([1,2,3])
t=tf.add(t_1,t_2)
with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    print(sess.run(t))

结果：runfile(‘D:/anaconda3/1/untitled1.py’, wdir=‘D:/anaconda3/1’)
[5 5 5]
很好，也很顺利。
运行乘法：multiply 没问题

import numpy as np
import tensorflow as tf
t_2 = tf.constant([4,3,1])
t_1 = tf.constant([[1],[2],[3]])
#t=tf.matmul(t_1,t_2)
t1=tf.multiply(t_1,t_2)
with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    print(sess.run(t1))
  

[[ 4 3 1]
[ 8 6 2]
[12 9 3]]
#运行matmul，直接傻了

import numpy as np
import tensorflow as tf
t_2 = tf.constant([4,3,1])
t_1 = tf.constant([[1],[2],[3]])
t=tf.matmul(t_1,t_2)
#t1=tf.multiply(t_1,t_2)
with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    print(sess.run(t))

ValueError: Shape must be rank 2 but is rank 1 for ‘MatMul_25’ (op: ‘MatMul’) with input shapes: [3,1], [3].
查了半天没搞明白，直接说格式对。后来发现t_2定义矩阵不对，改成
t_2 = tf.constant([[4,3,1]])搞定

import numpy as np
import tensorflow as tf
t_2 = tf.constant([[4,3,1]])
t_1 = tf.constant([[1],[2],[3]])
t=tf.matmul(t_1,t_2)
#t1=tf.multiply(t_1,t_2)
with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    print(sess.run(t))

[[ 4 3 1]
[ 8 6 2]
[12 9 3]]
基础不牢的代价！！