#!/usr/bin/python3
# -*- coding: utf-8 -*-
# 　

import numpy as np
import matplotlib.pyplot as plt
import tensorflow as tf
import sys
import time
import specrep

t = time.localtime()
print('executed at {:4}/{:02}/{:02} {:02}:{:02}:{:02}'.format(t.tm_year,t.tm_mon,t.tm_mday,t.tm_hour,t.tm_min,t.tm_sec))
print('python version:', sys.version)
print('numpy version:', np.__version__)
print('tensorflow version: ', tf.__version__)
if (not tf.__version__.startswith('1.')):
    print('This code requires version 1.*.')
    exit()


print('\nPrepare the training dateset.')
train_t = np.array([5.2, 5.7, 8.6, 14.9, 18.2, 20.4, 25.5, 26.4, 22.8, 17.5, 11.1, 6.6])
train_t = train_t.reshape([12, 1])
train_x = np.array([[mon**n for n in range(0, 5)]
                    for mon in range(1, 13)])
print('train_x =', train_x)
print('train_t =', train_t)
# x を5次元の入力データとみなし、12組の(x,t)のサンプルと理解する

print('\nDefine a placeholder to hold the training dataset.')
x = tf.compat.v1.placeholder(tf.float32, [None, 5])
print('x =', x)


print('\nDefine variables for weights.')
w = tf.Variable(tf.zeros([5, 1]))
print('w =', w)

print('\nDefine a ploynomial to predict label values.')
y = tf.matmul(x, w)
print('y =', y)
print('y.op.inputs._inputs[0] is x:', y.op.inputs._inputs[0] is x)
print('y.op.inputs._inputs[1] is w:', y.op.inputs._inputs[1] is w)
print('y.op.inputs._inputs[1].op.inputs[0] is w:', y.op.inputs._inputs[1].op.inputs[0] is w)
# 行列の乗算を実行しているわけではなく、式を登録している

print('\nDefine a placeholder to hold the label values.')
t = tf.compat.v1.placeholder(tf.float32, [None, 1])
print('t =', t)

print('\nDefine a loss function as Square error.')
loss = tf.reduce_sum(tf.square(y-t))
print('loss =', loss)

print('\nDefine an optimization step using the Adam optimizer')
train_step = tf.compat.v1.train.AdamOptimizer().minimize(loss)
print('train_step =', train_step)

print('\nCreate a session and initialize variables.')
sess = tf.compat.v1.Session()
sess.run(tf.compat.v1.global_variables_initializer())
# 実行結果 = session.run(コマンド相当, param=....) で文脈依存


print('\nApply the optimization step for 100,000 epochs.')
time_start = time.perf_counter()
i = 0
for _ in range(100000):
    i += 1
    sess.run(train_step, feed_dict={x: train_x, t: train_t})
    if i % 10000 == 0:
        loss_val = sess.run(loss, feed_dict={x: train_x, t: train_t})
        print('Step: {}, Loss: {}'.format(i, loss_val))
time_end = time.perf_counter()
print('time elapsed =', time_end-time_start)

print('\nShow weights after the training.')
ws = sess.run(w)
print('ws =', ws)

print('\nDefine a function to predict values using the final weights.')
def predict(x):
    pred = sum([ws[n][0] * x**n for n in range(0, 5)])
    return pred


print('\nPlot a chart for predictions.')
fig = plt.figure()
subplot = fig.add_subplot(1, 1, 1)
subplot.set_xlim(1, 12)
subplot.set_xticks(range(1, 13))
subplot.scatter(range(1, 13), train_t)
xs = np.linspace(1, 12, 100)
ys = predict(xs)
subplot.plot(xs, ys)
plt.savefig('ex1-1.png')
# plt.show()

print('\nLog of computer spec.')
specrep.report()