#!/usr/bin/env python
# coding: utf-8

# # Tensor Manipulation

# In[ ]:


# https://www.tensorflow.org/api_guides/python/array_ops
import tensorflow as tf
import numpy as np
import pprint
tf.set_random_seed(777)  # for reproducibility

pp = pprint.PrettyPrinter(indent=4)
sess = tf.InteractiveSession()


# ## Simple Array

# In[55]:


t = np.array([0., 1., 2., 3., 4., 5., 6.])
pp.pprint(t)
print(t.ndim) # rank
print(t.shape) # shape
print(t[0], t[1], t[-1])
print(t[2:5], t[4:-1])
print(t[:2], t[3:])


# # 2D Array

# In[58]:


t = np.array([[1., 2., 3.], [4., 5., 6.], [7., 8., 9.], [10., 11., 12.]])
pp.pprint(t)
print(t.ndim) # rank
print(t.shape) # shape


# ## Shape, Rank, Axis

# In[5]:


t = tf.constant([1,2,3,4])
tf.shape(t).eval()


# In[6]:


t = tf.constant([[1,2],
                 [3,4]])
tf.shape(t).eval()


# In[60]:


t = tf.constant([[[[1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12]],[[13, 14, 15, 16], [17, 18, 19, 20], [21, 22, 23, 24]]]])
tf.shape(t).eval()


# In[8]:


[
    [
        [
            [1,2,3,4], 
            [5,6,7,8],
            [9,10,11,12]
        ],
        [
            [13,14,15,16],
            [17,18,19,20], 
            [21,22,23,24]
        ]
    ]
]


# ## Matmul VS multiply

# In[9]:


matrix1 = tf.constant([[3., 3.]])
matrix2 = tf.constant([[2.],[2.]])
tf.matmul(matrix1, matrix2).eval()


# In[10]:


(matrix1*matrix2).eval()


# ## Watch out broadcasting

# In[11]:


matrix1 = tf.constant([[3., 3.]])
matrix2 = tf.constant([[2.],[2.]])
(matrix1+matrix2).eval()


# In[12]:


matrix1 = tf.constant([[3., 3.]])
matrix2 = tf.constant([[2., 2.]])
(matrix1+matrix2).eval()


# ## Random values for variable initializations 

# In[2]:


tf.random_normal([3]).eval()


# In[3]:


tf.random_uniform([2]).eval()


# In[4]:


tf.random_uniform([2, 3]).eval()


# ## Reduce Mean/Sum

# In[13]:


tf.reduce_mean([1, 2], axis=0).eval()


# In[14]:


x = [[1., 2.],
     [3., 4.]]


tf.reduce_mean(x).eval()


# In[15]:


tf.reduce_mean(x, axis=0).eval()


# In[16]:


tf.reduce_mean(x, axis=1).eval()


# In[17]:


tf.reduce_mean(x, axis=-1).eval()


# In[18]:


tf.reduce_sum(x).eval()


# In[19]:


tf.reduce_sum(x, axis=0).eval()


# In[20]:


tf.reduce_sum(x, axis=-1).eval()


# In[21]:


tf.reduce_mean(tf.reduce_sum(x, axis=-1)).eval()


# ## Argmax with axis

# In[22]:


x = [[0, 1, 2],
     [2, 1, 0]]
tf.argmax(x, axis=0).eval()


# In[23]:


tf.argmax(x, axis=1).eval()


# In[24]:


tf.argmax(x, axis=-1).eval()


# ## Reshape, squeeze, expand_dims

# In[25]:


t = np.array([[[0, 1, 2], 
               [3, 4, 5]],
              
              [[6, 7, 8], 
               [9, 10, 11]]])
t.shape


# In[26]:


tf.reshape(t, shape=[-1, 3]).eval()


# In[27]:


tf.reshape(t, shape=[-1, 1, 3]).eval()


# In[28]:


tf.squeeze([[0], [1], [2]]).eval()


# In[29]:


tf.expand_dims([0, 1, 2], 1).eval()


# ## One hot

# In[30]:


tf.one_hot([[0], [1], [2], [0]], depth=3).eval()


# In[31]:


t = tf.one_hot([[0], [1], [2], [0]], depth=3)
tf.reshape(t, shape=[-1, 3]).eval()


# ## casting

# In[32]:


tf.cast([1.8, 2.2, 3.3, 4.9], tf.int32).eval()


# In[33]:


tf.cast([True, False, 1 == 1, 0 == 1], tf.int32).eval()


# ## Stack

# In[34]:


x = [1, 4]
y = [2, 5]
z = [3, 6]

# Pack along first dim.
tf.stack([x, y, z]).eval()


# In[35]:


tf.stack([x, y, z], axis=1).eval()


# ## Ones like and Zeros like

# In[36]:


x = [[0, 1, 2],
     [2, 1, 0]]

tf.ones_like(x).eval()


# In[37]:


tf.zeros_like(x).eval()


# ## Zip
# 

# In[38]:


for x, y in zip([1, 2, 3], [4, 5, 6]):
    print(x, y)


# In[39]:


for x, y, z in zip([1, 2, 3], [4, 5, 6], [7, 8, 9]):
    print(x, y, z)


# ## Transpose

# In[40]:


t = np.array([[[0, 1, 2], [3, 4, 5]], [[6, 7, 8], [9, 10, 11]]])
pp.pprint(t.shape)
pp.pprint(t)


# In[41]:


t1 = tf.transpose(t, [1, 0, 2])
pp.pprint(sess.run(t1).shape)
pp.pprint(sess.run(t1))


# In[42]:


t = tf.transpose(t1, [1, 0, 2])
pp.pprint(sess.run(t).shape)
pp.pprint(sess.run(t))


# In[43]:


t2 = tf.transpose(t, [1, 2, 0])
pp.pprint(sess.run(t2).shape)
pp.pprint(sess.run(t2))


# In[44]:


t = tf.transpose(t2, [2, 0, 1])
pp.pprint(sess.run(t).shape)
pp.pprint(sess.run(t))