# coding:utf8
"""
author:无米的快乐生活~
"""
from __future__ import print_function
import tensorflow as tf

# 常量节点1
node1 = tf.constant(3.0, dtype=tf.float32)
# 常量节点2
node2 = tf.constant(4.0)

print(node1)
print(node2)
#Tensor("Const:0", shape=(), dtype=float32)
#Tensor("Const_1:0", shape=(), dtype=float32)

# 创建 Session 对象，并调用 run 函数执行节点
sess = tf.Session()
print(sess.run([node1, node2]))
#[3.0, 4.0]

node3 = tf.add(node1, node2)
print("node3:", node3)
#node3: Tensor("Add:0", shape=(), dtype=float32)
print("sess.run(node3):", sess.run(node3))
#sess.run(node3): 7.0

# 占位符，参数化接受外部输入
a = tf.placeholder(tf.float32)
b = tf.placeholder(tf.float32)
adder_node = a + b
print(adder_node)
#Tensor("add:0", dtype=float32)
print(tf.add(a,b))
#Tensor("Add_1:0", dtype=float32)

print(sess.run(adder_node, {a:3, b:4.5}))
#[ 3.  7.]
print(sess.run(adder_node, {a:[1,3], b:[2,4]}))
#[ 3.  7.]

add_and_tripple = adder_node * 3
print(sess.run(add_and_tripple, {a:3, b:4.5}))
#22.5



'''
创建并使用模型
'''
# 变量允许我们像graph 中添加可训练的参数，通过类型和初始value 创建
W = tf.Variable([.3], dtype=tf.float32)
b = tf.Variable([-.3], dtype=tf.float32)
x = tf.placeholder(tf.float32)
linear_model = W * x + b

print(W, b, x, linear_model)
#<tf.Variable 'Variable:0' shape=(1,) dtype=float32_ref> <tf.Variable 'Variable_1:0' shape=(1,) dtype=float32_ref> Tensor("Placeholder_2:0", dtype=float32) Tensor("add_1:0", dtype=float32)

#显式调用初始化所有变量的函数
init = tf.global_variables_initializer()
print(W, b, x, linear_model)
#<tf.Variable 'Variable:0' shape=(1,) dtype=float32_ref> <tf.Variable 'Variable_1:0' shape=(1,) dtype=float32_ref> Tensor("Placeholder_2:0", dtype=float32) Tensor("add_1:0", dtype=float32)
sess.run(init)
print(W, b, x, linear_model)
#<tf.Variable 'Variable:0' shape=(1,) dtype=float32_ref> <tf.Variable 'Variable_1:0' shape=(1,) dtype=float32_ref> Tensor("Placeholder_2:0", dtype=float32) Tensor("add_1:0", dtype=float32)


print(sess.run(linear_model, {x:[1, 2,3, 4]}))
#[ 0.          0.30000001  0.60000002  0.90000004]



'''
评估模型
'''

y = tf.placeholder(tf.float32)
squared_deltas = tf.square(linear_model - y)
loss = tf.reduce_sum(squared_deltas)
print(sess.run(loss, {x: [1, 2, 3, 4], y: [0, -1, -2, -3]}))
#23.66

# 我们非常精准地猜出了 W 和 b 的精确值……
fixW = tf.assign(W, [-1.])
fixb = tf.assign(b, [1.])
sess.run([fixW, fixb])

print(sess.run(loss, {x: [1, 2, 3, 4], y: [0, -1, -2, -3]}))
#0.0

'''
优化
'''
# 梯度下降优化，为什么是0.01？
optimizer = tf.train.GradientDescentOptimizer(0.01)
train = optimizer.minimize(loss)

# 从新初始化参数
sess.run(init)
for i in range(1000):
    sess.run(train, {x: [1, 2, 3, 4], y: [ 0, -1, -2, -3]})
print(sess.run([W, b]))
#[array([-0.9999969], dtype=float32), array([ 0.99999082], dtype=float32)]
print(sess.run([W, b, loss], {x: [1, 2, 3, 4], y: [ 0, -1, -2, -3]}))
#[array([-0.9999969], dtype=float32), array([ 0.99999082], dtype=float32), 5.6999738e-11]