Step03-Connecting Functions

 The goal of step 3 is to implement another function and combine multiple functions for calculation. 

First, let's implement Exp function.

class Exp(Function):

    def forward(self, x):

        return np.exp(x)

If we want to calculate $y=(e^{x^{2}})^{2}$, the code will look like:

A=Square()

B=Exp()

C=Square()

x=Variable(np.array(0.5))

a=A(x)

b=B(a)

y=C(b)

print(y.data)

Like $y=(e^{x^{2}})^{2}$, a function that is composed with several functions is called "Composite Function". Each elementry functions of composite function are simple, but we can execute more comlex computation by a seqence of several elementry functions.

We can describe the code above as a computational graph:

x -> A -> a -> B -> b -> C -> y

The reason why we describe the sequence of computation with computational graph is because it makes efficient to calculate the gradient of each variable. The algorithm of calculating gradient of each variable if "backpropagation". We will talk about this later.