simple_computations.py

#!/usr/bin/env python

import tensorflow as tf

a = tf.constant(1)
b = tf.constant(2)

c = a + b
d = a * b

V1 = tf.constant([1.,2.])
V2 = tf.constant([3.,4.])
M = tf.constant([[1.,2.]]) # Matrix, 2d
N = tf.constant([[1.,2.],[3.,4.]]) # Matrix, 2d
K = tf.constant([[[1.,2.],[3.,4.]]]) #Tensor, 3d+

# You can also compute on tensors
V3 = V1 + V2

# Operations are element-wise by default
M2 = M * M

# True matrix multiplication requires a Special Call
NN = tf.matmul(N,N)

# The above code only defines a TF "graph"
# Nothing has been computed yet
# For that, you first need to create a TF "session"

sess = tf.Session()

output = sess.run(NN)

print("NN is:")
print(output)

sess.close()

sess = tf.InteractiveSession()
print("M2 is:")
print(M2.eval())

# TF "variables" can change value
# useful for updating model weights
W = tf.Variable(0, name="weight")

# But variables must be initialized by TF before use
init_op = tf.initialize_all_variables()
sess.run(init_op)

print("W is:")
print(W.eval())

W += a
print("W after adding a:")
print(W.eval())

W += a
print("W after adding a again:")
print(W.eval())

# You can return or supply arbitrary nodes,
# i.e. check an intermediate value or
# sub your value in the middle of a computation

E = d + b # 1*2 + 2 = 4

print("E as defined")
print(E.eval())

# Let's see what d was at the same time
print("E and d:")
print(sess.run([E,d]))

# Use a custom d by specifying a dictionary
print("E with custom d=4")
print(sess.run(E, feed_dict = {d: 4}))