2.3.1. Forward differences (Cont.

Thus, we have

∆y0 = y1 − y 0

∆2 y0 = ∆(∆y0 ) = ∆y1 − ∆y0 = y2 − 2 y1 + y0 (2.3.1.4)

∆3 y0 = ∆(∆2 y0 ) = ∆2 y1 − ∆2 y0 = y3 − 3 y2 + 3 y1 − y0

∆4 y0 = ∆(∆3 y0 ) = ∆3 y1 − ∆3 y0 = y4 − 4 y3 + 6 y2 − 4 y1 + y0

and so on.

In general,

rk
k
∆k yi = ∑ (−1)  r  y
r =0
i+ k −r , (2.3.1.5)

where i = 0,1,..., n − 1 and k ( k = 1,..., n) is a positive integer.

2.3.1.1. Forward difference table

We can calculate the above forward differences very easily with the help of the following table

which is called forward difference table. It has been shown in Table1.

1
x y ∆y ∆2 y ∆3 y ∆4 y ∆5 y

x0 y0

∆y 0
x1 y1 ∆2 y0
∆ y1 ∆3 y0

x2 y2 ∆2 y1 ∆4 y0

∆3 y1
2
∆5 y0
∆y 2 ∆ y2
x3 y3 ∆4 y1
3
∆ y2
∆y 3 2
∆ y3
x4 y4

∆y 4
x5 y5

Table1 Forward difference table

2
2.4 The Shift, differentiation and averaging operators

2.4.1. The Shift operator

The shift operator or shifting operator E is defined by

Ef ( x) = f ( x + h ) (2.4.1.1)

Now, we know that

∆f ( x) = f ( x + h) − f ( x)

= Ef ( x) − f ( x)

= ( E − I ) f ( x) , where I is the identity operator.

So, it follows that

∆= E−I (2.4.1.2)

The eq. (2.4.1.2) represents a relation between the shift operator and the forward difference

operator ∆ .

Now, we have

Ef ( x) = f ( x + h)

E 2 f ( x ) = E .Ef ( x ) = Ef ( x + h ) = f ( x + 2 h )

E 3 f ( x ) = E . E 2 f ( x ) = Ef ( x + 2 h ) = f ( x + 3h )

3
By the method of induction, it may be easily proved that

E n f ( x ) = f ( x + nh ) , (2.4.1.3)

where n is a positive integer.

2.4.3. The averaging operator

The averaging operator µ is defined by

1  h  h 
µf ( x ) =  f  x +  + f  x −  (2.4.3.1)
2  2  2 

1 1/ 2
=
2
[
E f ( x ) + E −1 / 2 f ( x ) ]

1 1/ 2
=
2
[
E + E −1/ 2 f ( x)]
This implies that

1 1/ 2
µ=
2
[
E + E −1 / 2 ] (2.4.3.2)

The eq. (2.4.3.2) represents a relation between the averaging operator and the shift operator E .

 ∆2  4
Exercise 1: Prove that   x = 12x 2 + 2 , where the interval of differencing being unity.

E

Exercise 2: Find the missing terms in the following table

x 0 1 2 3 4 5

f(x) 0 − 8 15 − 35

4
2.6. Backward differences

Let y0 , y1 ,..., yn be a given set of values of y corresponding to the equidistant values x0 , x1 ,..., xn of

x , i.e. xi = x0 + ih , i = 0,1,...,n . The differences y1 − y 0 , y 2 − y1 , …, yn − yn −1 are called first

backward differences if these are denoted by ∇ y1 , ∇y 2 , …, ∇y n respectively. Thus we have

∇y i = yi − y i−1 , i = 1,..., n . (2.6.1)

The operator ∇ is called first backward difference operator.

In general, the first backward difference operator is defined by

∇f (x) = f ( x) − f ( x − h) (2.6.2)

= f ( x ) − E −1 f ( x )

= ( I − E −1 ) f ( x ) , where I is the identity operator.

So, it follows that

∇ = I − E −1 (2.6.2)

The eq. (2.6.2) represents a relation between the shift operator and the backward difference

operator ∇ .