Chapter- Four

Operators in Numerical Analysis (Difference Operators)

Lot of operators are used in numerical analysis/computation. Some of the frequently
used operators; forward difference (∆), backward difference (𝛻), central difference
(𝛿), shift (𝐸) and mean (𝜇) are discussed in this chapter.
Let the function 𝑦 = 𝑓 (𝑥) be defined on the closed interval [𝑎, 𝑏] and let 𝑥0 , 𝑥1 , . . . , 𝑥𝑛
be the n values of 𝑥. Assumed that these values are equidistance, i.e. 𝑥𝑖 = 𝑥0 + 𝑖ℎ,
𝑖 = 0, 1, 2, . . . , 𝑛; ℎ is a called the difference of the interval or spacing size. When 𝑥 = 𝑥𝑖 , the value
of 𝑦 is denoted by 𝑦𝑖 and is defined by 𝑦𝑖 = 𝑓(𝑥𝑖 ). The values of 𝑥 and 𝑦 are called arguments and
entries respectively.
The calculus of finite differences deals with change that take place in each value of the function
due to the independent variable changes by finite jumps.

Remark: In the infinitesimal calculus the independent variable changes continuously in a given
interval. If we have the following consecutive values of 𝑥 differing by ℎ: then 𝑓(𝑥) will have
accordingly
1. The Shift Operator: E
Let 𝑦 = 𝑓(𝑥) be the function of 𝑥 and let 𝑥, 𝑥 + ℎ, 𝑥 + 2ℎ, …be the consecutive values of 𝑥.
Then We define the shift operator 𝐸 by:
𝐸𝑓(𝑥) = 𝑓(𝑥 + ℎ) ⇒ 𝐸𝑓(𝑥𝑖 ) = 𝑓(𝑥𝑖 + ℎ)

and is called the shift operator, or displacement or Translation operator.

and
𝐸 2 𝑓(𝑥𝑖 ) = 𝐸(𝐸𝑓(𝑥𝑖 ) = 𝐸(𝑓(𝑥𝑖 + ℎ)) = 𝑓(𝑥𝑖 + ℎ + ℎ) = 𝑓(𝑥𝑖 + 2ℎ)
𝐸 2 𝑓(𝑥𝑖 ) = 𝐸(𝐸 2 𝑓(𝑥𝑖 )) = 𝐸(𝑓(𝑥𝑖 + 2ℎ )) = 𝑓(𝑥𝑖 + 3ℎ)
In general
𝐸 𝑛 𝑓(𝑥) = 𝑓( 𝑥 + 𝑛ℎ) ⇒ 𝐸 𝑛 𝑓(𝑥𝑖 ) = 𝑓( 𝑥𝑖 + 𝑛ℎ)
The inverse Shift Operator: 𝐄−𝟏
Let 𝑦 = 𝑓(𝑥) be the function of 𝑥 and let 𝑥, 𝑥 + ℎ, 𝑥 + 2ℎ, … be the consecutive values of 𝑥.
Then We define the shift operator 𝐸 by:
𝐸 −1 𝑓(𝑥)𝑓(𝑥 − ℎ) ⇒ 𝐸 −1 𝑓(𝑥𝑖 ) = 𝑓(𝑥𝑖 − ℎ)

1
and is called the inverse shift operator.
and
𝐸 −2 𝑓(𝑥𝑖 ) = 𝑓(𝑥𝑖 − 2ℎ)
In general, we have
𝐸 −𝑛 𝑓(𝑥𝑖 ) = 𝑓(𝑥𝑖 − 𝑛ℎ).
Remark
1. 𝐸 −1 𝐸 = 𝐸 0 = 𝐼 = 1 is the identity operator.
Properties
i. 𝐸[𝑐] = 𝑐, where 𝑐 is constant
ii. 𝐸[𝑐𝑓(𝑥)] = 𝑐𝐸𝑓(𝑥)
iii. 𝐸[𝑐1 𝑓1 (𝑥) + 𝑐2 𝑓2 (𝑥) + ⋯ + 𝑐𝑛 𝑓𝑛 (𝑥)] = 𝑐1 𝐸𝑓1 (𝑥) + 𝑐2 𝐸𝑓2 (𝑥) + ⋯ + 𝑐𝑛 𝐸𝑓𝑛 (𝑥)
iv. 𝐸 𝑚 [𝐸 𝑛 𝑓(𝑥) = 𝐸 𝑚+𝑛 𝑓(𝑥)
v. 𝐸 𝑚 [𝐸 −𝑚 𝑓(𝑥)] = 𝑓(𝑥)
vi. 𝐸 𝑚 [𝐸 𝑛 𝑓(𝑥)] = 𝐸 𝑚𝑛 [𝑓(𝑥)]
𝑓(𝑥) 𝐸𝑓(𝑥)
vii. 𝐸 [𝑔(𝑥)] = 𝐸𝑔(𝑥)

viii. 𝐸[𝑓(𝑥)𝑔(𝑥)] = 𝐸𝑓(𝑥)𝐸𝑔(𝑥)

Forward difference operator:∆

Suppose that a function 𝑓(𝑥) is given at equally spaced discrete points

say 𝑥0 , 𝑥1 , ⋯ , 𝑥𝑛 as 𝑓0 , 𝑓1 , ⋯ , 𝑓𝑛 respectively. Also let the constant difference between two
consecutive points of 𝑥 is called the interval of differencing or the step length denoted by ℎ. Then
the forward difference operator ∆ is defined as

∆𝑓(𝑥) = 𝑓(𝑥 + ℎ) − 𝑓(𝑥) ⇒ ∆𝑓𝑖 = 𝑓𝑖+1 − 𝑓𝑖

where any typical 𝑓𝑖 = 𝑓(𝑥𝑖 ).

Similarly, the higher differences are defined as

∆2 𝑓(𝑥) = ∆(∆𝑓(𝑥)) = ∆(𝑓(𝑥 + ℎ) − 𝑓(𝑥)) = 𝑓(𝑥 + 2ℎ) − 2𝑓(𝑥 + ℎ) + 𝑓(𝑥)

= 𝑓𝑖+2 − 2𝑓𝑖+1 + 𝑓𝑖

2
or in general,

∆𝑛 𝑓(𝑥) = ∆𝑛−1 (∆𝑓(𝑥)) = ∆𝑛−1 𝑓(𝑥 + ℎ) − ∆𝑛−1 𝑓(𝑥)

Forward – Difference Table

𝒙 𝒇(𝒙) ∆𝒇 ∆𝟐 𝒇 ∆𝟑 𝒇 ∆𝟒 𝒇 ∆𝟓 𝒇
𝒙𝟎 𝑓0
∆𝑓0
𝒙𝟏 𝑓1 ∆2 𝑓0
∆𝑓1 ∆3 𝑓0
𝒙𝟐 𝑓2 ∆2 𝑓1 ∆4 𝑓0
∆𝑓2 ∆3 𝑓1 ∆5 𝑓0
𝒙𝟑 𝑓3 ∆2 𝑓2 ∆4 𝑓1
∆𝑓3 ∆3 𝑓2
𝒙𝟒 𝑓4 ∆2 𝑓3
∆𝑓4

Example:

Let a function 𝑓(𝑥) is given at the points (0, 7), (4, 43), (8, 367) then find the forward difference
of the function at 𝑥 = 4.
Solution:
∆𝑓(𝑥) = 𝑓(𝑥 + ℎ) = 𝑓(𝑥)
⇒ ∆𝑓(4) = 𝑓(4 + ℎ) = 𝑓(4) ⇒ 𝑓(8) − 𝑓(4) = 367 − 43 = 324.

Example:

Find ∆𝑓(𝑥), ∆2 𝑓(𝑥) and ∆3 𝑓(𝑥) for the function 𝑥 2 + 2𝑥 + 3 with ℎ = 2.

Solution:
i. ∆𝑓(𝑥) = 𝑓(𝑥 + ℎ) = 𝑓(𝑥) = (𝑥 + 2)2 + 2(𝑥 + 2) + 3 − 𝑥 2 − 2𝑥 − 3
= 𝑥 2 + 4𝑥 + 4 + 2𝑥 + 4 + 3 − 𝑥 2 − 2𝑥 − 3 = 4𝑥 + 8.
ii. ∆2 𝑓(𝑥) = 4(𝑥 + 2) + 8 − 4𝑥 − 8 = 4𝑥 + 8 − 4𝑥 = 8.
iii. ∆3 𝑓(𝑥) = 8 − 8 = 0.

3
Properties:

1. ∆𝑐 = 0, where 𝑐 is a constant
2. ∆[𝑐𝑓 (𝑥)] = 𝑐∆𝑓 (𝑥)
3. ∆[𝑐1 𝑓1 (𝑥) + 𝑐2 𝑓2 (𝑥) + ⋯ + 𝑐𝑛 𝑓𝑛 (𝑥)] = 𝑐1 ∆𝑓1 (𝑥) + 𝑐2 ∆𝑓2 (𝑥) + ⋯ + 𝑐𝑛 ∆𝑓𝑛 (𝑥)

4. ∆(𝑓(𝑥) + 𝑔(𝑥)) = ∆𝑓(𝑥) + ∆𝑔(𝑥)

5. ∆(𝑐 𝑓(𝑥)) = ∆𝑓(𝑥) where 𝑐 is a constant
6. ∆[𝑓(𝑥)𝑔(𝑥)] = 𝑓(𝑥)∆𝑔(𝑥) + 𝑔(𝑥 + ℎ) ∆𝑓(𝑥)
𝑓(𝑥) 𝑔(𝑥)𝛥𝑓(𝑥)−𝑓(𝑥)𝛥𝑔(𝑥)
7. ∆ [𝑔(𝑥)] = 𝑔(𝑥)𝛥𝑔(𝑥)

8. ∆[𝑐 𝑥 ] = 𝑐 𝑥+ℎ − 𝑐 𝑥
9. If 𝑓(𝑥) is a polynomial of degree 𝑛,
𝑛

𝑓(𝑥) = ∑ 𝑎𝑖 𝑥 𝑖
𝑖=0

, then ∆𝑛 𝑓(𝑥) is constant and

∆𝑛 𝑓(𝑥) = 𝑛! 𝑎𝑛 ℎ𝑛
Proof:
6.
∆[𝑓 (𝑥)𝑔(𝑥)] = 𝑓(𝑥 + ℎ)𝑔(𝑥 + ℎ) − 𝑓(𝑥)𝑔(𝑥)
= 𝑓(𝑥 + ℎ)𝑔(𝑥 + ℎ) − 𝑓(𝑥 + ℎ)𝑔(𝑥) + 𝑓(𝑥 + ℎ)𝑔(𝑥) − 𝑓(𝑥)𝑔(𝑥)
= 𝑓 (𝑥 + ℎ)[𝑔(𝑥 + ℎ) − 𝑔(𝑥)] + 𝑔(𝑥)[𝑓 (𝑥 + ℎ) − 𝑓(𝑥)]
= 𝑓 (𝑥 + ℎ)∆𝑔(𝑥) + 𝑔(𝑥)∆𝑓(𝑥)
7.
𝑓(𝑥) 𝑓(𝑥 + ℎ) 𝑓(𝑥) 𝑓(𝑥 + ℎ)𝑔(𝑥) − 𝑔(𝑥 + ℎ)𝑓(𝑥)
∆[ ] = − =
𝑔(𝑥) 𝑔(𝑥 + ℎ) 𝑔(𝑥) 𝑔(𝑥 + ℎ)𝑔(𝑥)
𝑓 (𝑥 + ℎ)𝑔(𝑥) − 𝑓 𝑥 𝑔 𝑥 + 𝑓(𝑥)𝑔(𝑥) − 𝑔(𝑥 + ℎ)𝑓(𝑥)
( ) ( )
=
𝑔(𝑥 + ℎ)𝑔(𝑥)
𝑔(𝑥)[𝑓 (𝑥 + ℎ) − 𝑓(𝑥)] − 𝑓(𝑥)[𝑔(𝑥 + ℎ) − 𝑔(𝑥)]
=
𝑔(𝑥 + ℎ)𝑔(𝑥)
𝑔(𝑥)𝛥𝑓(𝑥) − 𝑓(𝑥)𝛥𝑔(𝑥)
= .
𝑔(𝑥)𝛥𝑔(𝑥)
In particular
1 𝛥𝑓(𝑥))
∆[ ]=− .
𝑓(𝑥) 𝑓(𝑥 + ℎ)𝑓(𝑥)

4
 9. Let’s prove this by mathematical induction for 𝑛 = 1,
𝑓(𝑥) = 𝑎0 + 𝑎1 𝑥 and
∆𝑓(𝑥) = 𝑎0 + 𝑎1 (𝑥 + ℎ) = 𝑎0 + 𝑎1 𝑥 = 𝑎1ℎ = 𝑎1 1! ℎ1
assume that the formula is true for all polynomials up to the degree 𝑛 − 1. Consider
the 𝑛𝑡ℎ degree polynomial
𝑛−1

𝑓(𝑥) = ∑ 𝑎𝑖 𝑥 𝑖
𝑖=0
𝑛

∆ 𝑓(𝑥) = ∆ (∑ 𝑎𝑖 𝑥 𝑖 )
𝑛 𝑛

𝑖=0
𝑛−1
𝑛
∆ 𝑓(𝑥) = ∆ (∑ 𝑎𝑖 𝑥 𝑖 ) + ∆𝑛 𝑓(𝑥) + ∆𝑛 𝑎𝑛 𝑥 𝑛 = 0 + 𝑎𝑛 ∆ 𝑥𝑛
𝑛 𝑛

𝑖=0

= 𝑎𝑛 (∆𝑛−1 (𝑥 + ℎ)𝑛 − ∆𝑛 𝑥𝑛 ) = 𝑎𝑛 ∆𝑛−1 ( 𝑛ℎ𝑥𝑛−1 𝑔(𝑥))

where 𝑔(𝑥) is a polynomial of degree less than (𝑛 − 1) and hence ∆𝑛−1 𝑔(𝑥) = 0.
⇒ ∆𝑛 𝑓(𝑥) = 𝑎𝑛 ∆𝑛−1 ( 𝑛ℎ𝑥 𝑛−1 ) = 𝑎𝑛 (𝑛ℎ)(𝑛 − 1)! ℎ𝑛−1 = 𝑎𝑛 𝑛! ℎ𝑛 .

Backward difference operator: ∇

The difference operator defined by:

∇𝑓(𝑥) = 𝑓(𝑥) − 𝑓(𝑥 − ℎ) ⇒ ∇𝑓𝑖 = 𝑓𝑖 − 𝑓𝑖−1

is called backward difference operator.

∇2 𝑓𝑖 = ∇(∇𝑓𝑖 ) = ∇(𝑓𝑖 − 𝑓𝑖−1 ) = 𝑓𝑖 − 2𝑓𝑖−1 + 𝑓𝑖−2 and so on.

5
Backward – Difference Table

𝒙 𝒇(𝒙) 𝛁𝒇 𝛁𝟐𝒇 𝛁𝟑𝒇 𝛁𝟒𝒇 𝛁𝟓𝒇

𝒙𝟎 𝑓0
∇𝑓1
𝒙𝟏 𝑓1 ∇2 𝑓2
∇𝑓2 ∇3 𝑓3
𝒙𝟐 𝑓2 ∇2 𝑓3 ∇4 𝑓4
∇𝑓3 ∇3 𝑓4 ∇5 𝑓5
𝒙𝟑 𝑓3 ∇2 𝑓4 ∇4 𝑓5
∇𝑓4 ∇3 𝑓5
𝒙𝟒 𝑓4 ∇2 𝑓5
∇𝑓5
𝒙𝟓 𝑓5

Error propagation in a difference table

If any entry of the difference table is erroneous, then this error spread over the table

in convex manner. The propagation of error in a difference table is illustrated in the following
table. Let us assumed that 𝑓3 be erroneous and the amount of the error be 𝜀. Following observations
are noted from the table.

6
 𝑥 𝑓(𝑥) ∆𝑓 ∆2 𝑓 ∆3 𝑓 ∆4 𝑓 ∆5 𝑓
𝑥0 𝑓0
∆𝑓0
𝑥1 𝑓1 ∆2 𝑓0
∆𝑓1 ∆3 𝑓0 + 𝜀
𝑥2 𝑓2 ∆2 𝑓1 + 𝜀 ∆4 𝑓0 − 4𝜀
∆𝑓2 + 𝜀 ∆3 𝑓1 − 3𝜀 ∆5 𝑓0 + 10𝜀
𝑥3 𝑓3 + 𝜀 ∆2 𝑓2 − 2𝜀 ∆4 𝑓1 + 6𝜀
∆𝑓3 − 𝜀 ∆3 𝑓2 + 3𝜀 ∆5 𝑓1 + 10𝜀
𝑥4 𝑓4 ∆2 𝑓3 + 𝜀 ∆4 𝑓2 − 4𝜀
∆𝑓4 ∆3 𝑓3 − 𝜀
𝑥5 𝑓5 ∆2 𝑓4
∆𝑓5
𝑥6 𝑓6

Table 1: Error propagation in a finite difference table.

i. The error increases with the order of the differences.

ii. The error is maximum (in magnitude) along the horizontal line through the erroneous
tabulated value.
iii. In the 𝑛𝑡ℎ difference column, the coefficients of errors are the binomial coefficients in
the expansion of (1 − 𝜀)𝑛 . In particular, the errors in the second difference column are
𝜀, −2𝜀, 𝜀, in the third difference column these are 𝜀, −3𝜀, 3𝜀, −𝜀, in the fourth difference
column are 𝜀, − 4𝜀, 6𝜀, − 4𝜀, 𝜀 and so on.
iv. The algebraic sum of errors in any complete column is zero.
v. The maximum error in each column, occurs opposite to the entry containing the error, i.e., 𝑓5.

If there is any error in a single entry of the table, then we can detect and correct it from the
difference table. The position of the error in an entry can be identified by performing the following
steps.

i. If at any stage, the differences do not follow a smooth pattern, then there is an error.

7
 ii. the differences of some order (it is generally happening in higher order) becomes
alternating in sign then the middle entry contains an error.

Central difference operator: 𝜹

Definition: The central difference operator is defined by

ℎ ℎ
𝑓(𝑥) = 𝑓 (𝑥 + ) – 𝑓 (𝑥 − )
2 2
= 𝑓1 − 𝑓−1
2 2
1 −1⁄
= (𝐸 ⁄2 −𝐸 2 ) 𝑓(𝑥)

−1⁄
=𝐸 2 (𝐸 – 1)𝑓(𝑥)
−1⁄
=𝐸 2 ∆𝑓(𝑥)

Central – Difference Table

𝒙 𝒇(𝒙) 𝛅𝒇 𝛅𝟐 𝒇 𝛅𝟑 𝒇 𝛅𝟒 𝒇 𝛅𝟓 𝒇
𝒙𝟎 𝑓0
δ𝑓1/2
𝒙𝟏 𝑓1 δ2 𝑓1
δ𝑓3/2 δ3 𝑓3/2
𝒙𝟐 𝑓2 δ2 𝑓2 δ4 𝑓2
δ𝑓5/2 δ3 𝑓5/2 δ5 𝑓5/2
𝒙𝟑 𝑓3 δ2 𝑓3 δ4 𝑓3
δ𝑓7/2 δ3 𝑓7/2
𝒙𝟒 𝑓4 δ2 𝑓4
δ𝑓9/2
𝒙𝟓 𝑓5

Averaging operator 𝝁:

Definition: the averaging operator 𝜇 defined

1 1
1 ℎ ℎ 1
𝜇𝑓(𝑥) = 2 [𝑓 (𝑥 + 2 ) + 𝑓 (𝑥 − 2)] = 2 [𝐸 2 + 𝐸 −2 𝑓] 𝑓(𝑥).

8
 Differential operator, 𝑫:

The differential operator is well known from differential calculus and it is denoted by

𝐷. This operator gives the derivative. That is,

𝑑
𝐷𝑓(𝑥) = 𝑓(𝑥) = 𝑓′(𝑥)
𝑑𝑥
𝑑2
𝐷2 𝑓(𝑥) = 𝑓(𝑥) = 𝑓"(𝑥)
𝑑𝑥2
⋮
𝑛
𝑑
𝐷 𝑛 𝑓 ( 𝑥) = 𝑓 ( 𝑥) = 𝑓 𝑛 ( 𝑥)
𝑑𝑥𝑛

Some relations between various operators:

1. ∆= 𝐸 − 1.
2. ∇= 1 − 𝐸 −1
1
3. ∆= 𝐸∇= ∇𝐸 = 𝛿𝐸 2
4. 𝐸 = 𝑒 ℎ𝐷
∆ ∇
5. −∆=∆+∇
∇

6. ∇= ∆𝐸 −1
𝛿4
7. 𝜇 = √1 + 4

8. ∆= 𝑒 ℎ𝐷 − 1
9. ∇∆= ∆∇= 𝛿 2
1 1
10. ∇(1 − ∇)−2 = ∆(1 − ∆)−2
11. ℎ𝐷 = sinh−1(𝜇𝛿).

Proof:

1. We have

∆𝑓(𝑥) = 𝑓(𝑥 + ℎ) − 𝑓(𝑥) = (𝐸 − 1)𝑓(𝑥)

∴ ∆= 𝐸 − 1

2. Since∇𝑓(𝑥) = 𝑓(𝑥) − 𝑓(𝑥 − ℎ), then

9
 ∆𝑓(𝑥) = (1 − 𝐸 −1 )𝑓(𝑥)
Therefore,
∇= (1 − 𝐸 −1 )

3. 𝐸∇𝑓(𝑥) = 𝐸(𝑓(𝑥) − 𝑓(𝑥 − ℎ)) = 𝑓(𝑥 + ℎ) − 𝑓(𝑥) = ∆𝑓(𝑥)

Hence,
𝐸∇= ∆

1 ℎ 1
𝛿𝐸 2 𝑓(𝑥) = 𝛿𝑓 (𝑥 + ) = 𝑓(𝑥 + ℎ) − 𝑓(𝑥) = ∆𝑓(𝑥) ⟹ 𝛿𝐸 2 = ∆
2
Therefore,
1
∆= 𝐸∇= ∇𝐸 = 𝛿𝐸 2

4. The Taylor series representation of 𝑓(𝑥 + ℎ) about 𝑥 is given as

′ (𝑥)
ℎ2 ′′ ℎ3 ′′′
𝑓(𝑥 + ℎ) = 𝑓(𝑥) + ℎ𝑓 + 𝑓 (𝑥) + 𝑓 (𝑥) + ⋯
2! 3!
The above equation can write as
ℎ2 2 ℎ3 3
𝐸𝑓(𝑥) = (1 + ℎ𝐷 + 𝐷 + 𝐷 + ⋯ ) 𝑓(𝑥) = 𝑒 ℎ𝐷 𝑓(𝑥)
2! 3!
Therefore,

𝐸 = 𝑒 ℎ𝐷
5.
∆ ∇ 𝐸−1 1 − 𝐸 −1 1
( − ) 𝑓(𝑥) = ( − ) 𝑓(𝑥) = (𝐸 − ) 𝑓(𝑥) = (𝐸 − 𝐸 −1 )𝑓(𝑥)
∇ ∆ 1 − 𝐸 −1 𝐸−1 𝐸
= ([𝐸 − 1] + [1 − 𝐸 −1 ])𝑓(𝑥) = (∆ + ∇)𝑓(𝑥)
Therefore,
∆ ∇
− =∆+∇
∇ ∆
6. ∇𝑓(𝑥) = 𝑓(𝑥) − 𝑓(𝑥 − ℎ) = ∆𝑓(𝑥 − ℎ) = ∆𝐸 −1 𝑓(𝑥)
Hence,

10
 ∇= ∆𝐸 −1
7.

1 1 2
−
√(𝐸 + 𝐸 )
2 2
1 ℎ ℎ 1 1 1
𝜇 = [𝑓 (𝑥 + ) + 𝑓 (𝑥 − )] = [𝐸 2 + 𝐸 −2 ] 𝑓(𝑥) = 𝑓(𝑥)
2 2 2 2 4

𝐸 + 𝐸 −1 + 2 𝐸 + 𝐸 −1 − 2 + 4
=√ 𝑓(𝑥) = √ 𝑓(𝑥)
4 4

1 1 2
𝐸 + 𝐸 −1 − 2 √ (𝐸 2 − 𝐸 −2 ) 𝛿2
= √1 + 𝑓(𝑥) = 1 + 𝑓(𝑥) = √1 + 𝑓(𝑥)
4 4 4

Hence,

𝛿4
𝜇 = √1 + .
4

8.

ℎ2 ′′ ℎ3
∆𝑓(𝑥) = 𝑓(𝑥 + ℎ) − 𝑓(𝑥) = 𝑓(𝑥) + ℎ𝑓 ′ (𝑥) + 𝑓 (𝑥) + 𝑓 ′′′ (𝑥) + ⋯ − 𝑓(𝑥)
2! 3!
ℎ2 2 ℎ3 3
= (1 + ℎ𝐷 + 𝐷 + 𝐷 + ⋯ ) 𝑓(𝑥) − 𝑓(𝑥)
2! 3!
= (𝑒 ℎ𝐷 − 1 )𝑓(𝑥)

Therefore,

∆= 𝑒 ℎ𝐷 − 1
9. From (1) and (2), we have
∆∇= (𝐸 − 1)(1 − 𝐸 −1 ) = (1 − 𝐸 −1 )(𝐸 − 1) = ∇∆
Hence,
∆∇= ∇∆
Again, we have
1 2 1 2 1 1 1 1 2
∆∇= (𝐸 − 1)(1 − 𝐸 −1 ) = ([𝐸 2 ] + [𝐸 −2 ] − 2𝐸 2 𝐸 −2 ) = (𝐸 2 − 𝐸 −2 ) = 𝛿 2

Therefore,

∇∆= ∆∇= 𝛿 2
11
 1 1
10. Since, 𝛿 = 𝐸 2 − 𝐸 −2 , we have
1 1 1 1 1
𝐸 2 − 𝐸 −2 = (1 − 𝐸 −1 )𝐸 2 = (1 − 𝐸 −1 )(𝐸 −1 )−2 = (1 − 𝐸 −1 )(1 − 1 + 𝐸 −1 )−2
1 1
= (1 − 𝐸 −1 )(1 − [1−𝐸 −1 ])−2 = (1 − 𝐸 −1 )(1 − [1−𝐸 −1 ])−2
1
= ∇(1 − ∇)−2 (1)
Again
1 1 1 1 1
𝐸 2 − 𝐸 −2 = (𝐸 − 1)𝐸 −2 = (𝐸 − 1)(𝐸 − 1 + 1)−2 = (𝐸 − 1)([𝐸 − 1] + 1)−2 =
1
= ∆(1 − ∆)−2 (2)
From (1) and (2), we obtain
1 1
𝛿 = ∇(1 − ∇)−2 = ∆(1 − ∆)−2
11.
1 1 ℎ𝐷 ℎ𝐷 ℎ𝐷
𝛿𝑓(𝑥) = [𝐸 2 − 𝐸 −2 ] 𝑓 (𝑥) = [𝑒 2 − 𝑒 − 2 ] 𝑓 (𝑥) = 2 sinh ( ) 𝑓(𝑥).
2
Thus,

ℎ𝐷 ℎ𝐷
𝛿 = 2 sinh ( ) , similarly 𝜇 = cosh ( ) .
2 2
Again,

ℎ𝐷 ℎ𝐷 ℎ𝐷
𝜇𝛿 = cosh ( ) . 2 sinh ( ) = sin h ( ) .
2 2 2
This relation gives the inverse result,

ℎ𝐷 = sinh−1(𝜇𝛿).

Note:

Since ℎ𝐷 = ln 𝐸 and using the relations 𝐸 = 1 + ∆ and 𝐸 −1 = 1 − 𝛻, we obtained,

ℎ𝐷 = ln 𝐸 = ln(1 + ∆ ) = − ln(1 − 𝛻) = sinh−1(𝜇𝛿) .

Remark

1. 𝜇𝐸𝑓 (𝑥) = 𝐸𝜇𝑓 (𝑥)

Example: Prove that the following relations

12
 1. (1 + ∆)(1 − ∇) = 1 ℎ𝐷
5. 𝜇 = cosh ( 2 )

∆+∇ ∆𝐸 −1 ∆
2. 𝜇𝛿 = 6. 𝜇𝛿 = +2
2 2

1 2
𝛿 𝛿2
3. 𝐸 2 = 𝜇 + 2 7. 1 + 𝛿 2 𝜇 2 = (1 + )
2

𝛿2 𝛿2
4. ∆= + 𝛿 √1 +
2 4

Proof:

1.

(1 + ∆)(1 − ∇)𝑓(𝑥) = (1 + ∆)(𝑓(𝑥) − 𝑓(𝑥) + 𝑓(𝑥 − ℎ)) = (1 + ∆)𝑓(𝑥 − ℎ)

= 𝑓(𝑥 − ℎ) − 𝑓(𝑥) + 𝑓(𝑥 − ℎ)
= 𝑓(𝑥)

Therefore,

(1 + ∆)(1 − ∇) = 1.

2.
∆+∇ 1 1
( ) 𝑓(𝑥) = (∆𝑓(𝑥) + ∇𝑓(𝑥)) = (𝑓(𝑥 + ℎ) − 𝑓(𝑥) + 𝑓(𝑥 − ℎ))
2 2 2
1
= (𝐸 − 𝐸 −1 )𝑓(𝑥) = 𝜇𝛿𝑓(𝑥).
2
Thus,

∆+∇
𝜇𝛿 = ( ).
2
1 1 1 1 1
𝛿 1 1
3. (𝜇 + 2) 𝑓(𝑥) = {2 [𝐸 2 + 𝐸 −2 ] + 2 [𝐸 2 − 𝐸 −2 ]} 𝑓(𝑥) = 𝐸 2 𝑓(𝑥).

Hence,
1 𝛿
𝐸2 = 𝜇 +
2
4.

13
 𝛿2 𝛿2 1 1 −
1 2 1
−
1 1 1 −
1 2
√ 2 2 2 √
[ + 𝛿 1 + ] 𝑓(𝑥) = [ (𝐸 − 𝐸 ) + (𝐸 − 𝐸 ) 1 + (𝐸 − 𝐸 2 ) ] 𝑓(𝑥)
2 2
2 4 2 4

1 1 1 1
= [ (𝐸 + 𝐸 −1 − 2) + (𝐸 2 − 𝐸 −2 ) √1 + (𝐸 + 𝐸 −1 − 2)] 𝑓(𝑥)
2 4

1 1 1 1
= [ (𝐸 + 𝐸 −1 − 2) + (𝐸 2 − 𝐸 −2 ) √ (𝐸 + 𝐸 −1 + 2)] 𝑓(𝑥)
2 4

1 1 1 1 1 1 2
= [ (𝐸 + 𝐸 −1 − 2) + (𝐸 2 − 𝐸 −2 ) √ (𝐸 2 + 𝐸 −2 ) ] 𝑓(𝑥)
2 4
1 1 1 1 1 1
= [ (𝐸 + 𝐸 −1 − 2) + (𝐸 2 − 𝐸 −2 ) (𝐸 2 + 𝐸 −2 )] 𝑓(𝑥)
2 2
1 1
= [ (𝐸 + 𝐸 −1 − 2) + (𝐸 − 𝐸 −1 )] 𝑓(𝑥)
2 2
= (𝐸 − 1)𝑓(𝑥)
Thus,

𝛿2 𝛿2
𝐸−1= + 𝛿 √1 + .
2 4

Example: Given that 𝑦0 = 3, 𝑦1 = 12, 𝑦2 = 81, 𝑦3 = 200, 𝑦4 = 100, 𝑦5 = 8, find the value of
∆5 𝑦0 without using difference table.

Solution: Using the relation between shift operator and forward difference operator, we have

∆5 𝑦0 = (𝐸 − 1)5 𝑦0 = 𝐸 5 𝑦0 − 5𝐸 4 𝑦0 + 10𝐸 3 𝑦0 − 10𝐸 2 𝑦0 + 5𝐸𝑦0 − 𝑦0

= 𝑦5 − 5𝑦4 + 10𝑦3 − 10𝐸 2 𝑦2 + 5𝑦1 − 𝑦0
= 8 − 5(100) + 10(200) − 10(81) + 5(12) − 3 = 755

Therefore, ∆5 𝑦0 = 755.

Divided Difference

Let (𝑥0 , 𝑦0 ), (𝑥1 , 𝑦1 ), (𝑥2 , 𝑦2 ), … , (𝑥𝑛 , 𝑦𝑛 ) be given (𝑛 + 1) points. Let 𝑦0 , 𝑦1 , 𝑦2 , … , 𝑦𝑛 be

the values of the function 𝑦 = 𝑓(𝑥) corresponding to the values of the argument
𝑥 = 𝑥0 , 𝑥1 , 𝑥2 , … , 𝑥𝑛 which are not necessarily equally spaced. Since the difference of the
function values with respect to the difference of the arguments are called divided differences.
Divided differences of order 0, 1, 2, … , 𝑘 are defined recursively by the relations

14
 1. 0𝑡ℎ order divided difference: 𝑓[𝑥0 ] = 𝑓(𝑥0 )
𝑓[𝑥1 ]−𝑓[𝑥0 ]
2. 1𝑠𝑡 order divided difference: 𝑓[𝑥0 , 𝑥0 ] = (𝑥1 −𝑥0 )

𝑓[𝑥1 ,𝑥2 ]−𝑓[𝑥0 ,𝑥1 ]

3. 2𝑛𝑑 order divided difference: 𝑓[𝑥0 , 𝑥1 , 𝑥2 ] = (𝑥2 −𝑥0 )
and

4. The 𝑘 𝑡ℎ order differences:

𝑓[𝑥1 , 𝑥2 , ⋯ , 𝑥𝑘 ] − 𝑓[𝑥0 , 𝑥1 , ⋯ , 𝑥𝑘−1 ]
𝑓[𝑥0 , 𝑥1 , 𝑥2 , … , 𝑥𝑘 ] =
(𝑥𝑘 − 𝑥0 )
Theorem: prove that
𝑘
𝑓(𝑥𝑗 )
𝑓[𝑥0 , 𝑥1 , 𝑥2 , … , 𝑥𝑘 ] = ∑ 𝑘
.
∏ (𝑥 − 𝑥𝑖 )
𝑗=0 𝑖=0 𝑗
𝑖≠𝑗

Proof: We will carry out the proof by induction for 𝑘 = 1

for k = 1,
𝑘
𝑓[𝑥1 ] − 𝑓[𝑥0 ] 𝑓(𝑥𝑗 )
𝑓[𝑥0 , 𝑥1 ] = 𝑓[𝑥0 , 𝑥0 ] = =∑ 1
(𝑥1 − 𝑥0 ) ∏𝑖=0(𝑥𝑗 − 𝑥𝑖 )
𝑗=0
𝑖≠𝑗

Suppose it is true for 𝑘  𝑟 ,then

𝑓[𝑥1 , ⋯ 𝑥𝑟+1 ] − 𝑓[𝑥0 , ⋯ , 𝑥𝑟 ]

𝑓[𝑥0 , 𝑥1 , … , 𝑥𝑟+1 ] =
(𝑥𝑟+1 − 𝑥0 )
𝑓[𝑥1 , 𝑥2 , ⋯ , 𝑥𝑟+1 ] − 𝑓[𝑥0 , 𝑥1 , ⋯ , 𝑥𝑟 ]
𝑓[𝑥0 , 𝑥1 , 𝑥2 , … , 𝑥𝑘 ] =
(𝑥𝑟+1 − 𝑥0 )

𝑟+1 𝑟
1 𝑓(𝑥𝑗 ) 𝑓(𝑥𝑗 )
= ∑ 𝑟+1 −∑ 𝑟
𝑥𝑟+1 − 𝑥0 ∏𝑖=1 (𝑥𝑗 − 𝑥𝑖 ) ∏𝑖=0(𝑥𝑗 − 𝑥𝑖 )
𝑗=1 𝑗=0
𝑖≠𝑗 𝑖≠𝑗
[ ]
𝑟 𝑟
1 𝑓(𝑥𝑟+1 ) 𝑓(𝑥𝑗 ) 𝑓(𝑥𝑗 ) 𝑓(𝑥0 )
= [ 𝑟 + ∑ 𝑟+1 −∑ 𝑟 − 𝑟 ]
𝑥𝑟+1 − 𝑥0 ∏𝑖=1(𝑥𝑟+1 − 𝑥𝑖 ) ∏𝑖=1 (𝑥𝑗 − 𝑥𝑖 ) ∏𝑖=0(𝑥𝑗 − 𝑥𝑖 ) ∏𝑖=1(𝑥0 − 𝑥𝑖 )
𝑗=1 𝑗=1 𝑖≠𝑗
𝑖≠𝑗

15
 𝑟
1 𝑓(𝑥𝑟+1 ) [(𝑥𝑗 − 𝑥0 ) − (𝑥𝑗 − 𝑥𝑟+1 )]
=( ) 𝑟 +∑ 𝑓(𝑥𝑗 )
𝑥𝑟+1 − 𝑥0 ∏𝑖=1(𝑥𝑟+1 − 𝑥𝑖 ) (𝑥𝑟+1 − 𝑥0 ) ∏𝑟+1
𝑖=1 (𝑥𝑗 − 𝑥𝑖 )
𝑗=1
𝑖≠𝑗
𝑟+1
𝑓(𝑥0 ) 𝑓(𝑥𝑗 )
− 𝑟 = ∑ 𝑟+1
(𝑥𝑟+1 − 𝑥0 ) ∏𝑖=1(𝑥0 − 𝑥𝑖 ) ∏𝑖=0 (𝑥𝑗 − 𝑥𝑖 )
𝑗=0
𝑙≠𝑗

Hence, the proof.

Divided Differences table
𝒙 𝒇(𝒙) 𝟏𝐬𝐭 DDD 𝟐𝒏𝒅 𝐃𝐃 𝟑𝒓𝒅 𝐃𝐃 𝟒𝒓𝒕𝒉 𝐃𝐃𝐃
𝒙𝟎 𝑓0
𝑓[𝑥0 , 𝑥1 ]
𝒙𝟏 𝑓1 𝑓[𝑥0 , 𝑥1 , 𝑥2 ]
𝑓[𝑥1 , 𝑥2 ] 𝑓[𝑥0 , 𝑥1 , 𝑥2 , 𝑥3 ]
𝒙𝟐 𝑓2 𝑓[𝑥1 , 𝑥2 , 𝑥3 ] 𝑓[𝑥0 , 𝑥1 , 𝑥2 , 𝑥3 , 𝑥4 ]
𝑓[𝑥2 , 𝑥3 ] 𝑓[𝑥1 , 𝑥2 , 𝑥3 , 𝑥4 ]
𝒙𝟑 𝑓3 𝑓[𝑥2 , 𝑥3 , 𝑥4 ]
𝑓[𝑥3 , 𝑥4 ]
𝒙𝟒 𝑓4

Properties of divided difference operator

❖ The divided differences are symmetrical in their arguments; that is, the value of any
divided difference is independent of the order of the arguments.
❖ The divided difference operator is linear.
❖ The 𝒏𝒕𝒉 order divided difference of a polynomial of degree 𝒏 is constant, equal to the
coefficient of 𝒙𝒏 .

1. Gregory-Newton Forward difference Interpolating Polynomial

Statement:
If 𝑦0 , 𝑦1 , 𝑦2 , … , 𝑦𝑛 be the values of the function 𝑦 = 𝑓(𝑥) corresponding to the equidistant
value of 𝑥 = 𝑥0 , 𝑥1 , 𝑥2 , … , 𝑥𝑛 , where 𝑥𝑖 − 𝑥𝑖−1 = ℎ, 𝑖 = 1, 2, … , 𝑛, then

16
 𝑢(𝑢 − 1) 2 𝑢(𝑢 − 1) … (𝑢 − (𝑛 − 1)) 𝑛
𝑦 = 𝑦0 + 𝑢∆𝑦0 + ∆ 𝑦0 + ⋯ + ∆ 𝑦0 (1)
2! 𝑛!
𝑥−𝑥0
, where 𝑢 = .
ℎ

Proof:
Since the points are 𝑛 + 1, the polynomial in 𝑥 used to approximate is in degree 𝑛.
Define the polynomial as:
𝑦 = 𝑎0 + 𝑎1 (𝑥 − 𝑥0 ) + 𝑎2 (𝑥 − 𝑥0 )(𝑥 − 𝑥1 ) + ⋯ + 𝑎𝑛 (𝑥 − 𝑥0 )(𝑥 − 𝑥1 ) … (𝑥 − 𝑥𝑛 ) (2)
But at 𝑥 = 𝑥𝑖 , 𝑦𝑖 = 𝑓(𝑥𝑖 ) = 𝑃𝑛 (𝑥𝑖 ).
If 𝑥 = 𝑥0 , 𝑎0 = 𝑦0
If 𝑥 = 𝑥1 ,
𝑦1 − 𝑦0
𝑦1 = 𝑦0 + 𝑎1 (𝑥1 − 𝑥0 ) ⇒ 𝑦1 − 𝑦0 = 𝑎1 (𝑥1 − 𝑥0 ) ⇒ = 𝑎1
ℎ
∆𝑦0
Therefore, 𝑎1 = .
ℎ

If 𝑥 = 𝑥2 ,
∆𝑦0
𝑦2 = 𝑦0 + (𝑥2 − 𝑥0 ) + 𝑎2 (𝑥2 − 𝑥0 )(𝑥2 − 𝑥1 ) ⇒ 𝑦2 = 𝑦0 + 2∆𝑦0 + 2ℎ2 𝑎2
ℎ
Therefore,
𝑦2 − 𝑦0 − 2∆𝑦0 ∆2 𝑦0
𝑎2 = = .
2ℎ2 2! ℎ2
Similarly,
∆3 𝑦0
𝑎3 = ,
3! ℎ3
∆𝑛 𝑦0
𝑎𝑛 = .
𝑛! ℎ𝑛
Substituting all value of 𝑎𝑖 ′s in (2) we get
𝑛
𝑢(𝑢 − 1) 2 𝑢(𝑢 − 1) … (𝑢 − (𝑛 − 1)) 𝑛 𝑢
𝑦 = 𝑦0 + 𝑢∆𝑦0 + ∆ 𝑦0 + ⋯ + ∆ 𝑦0 = ∑ ( ) ∆𝑘 𝑦0 (3)
2! 𝑛! 𝑘
𝑘=0

2. Gregory-Newton Backward Difference Interpolating Polynomial

Statement:
If 𝑦0 , 𝑦1 , 𝑦2 , … , 𝑦𝑛 be the values of the function 𝑦 = 𝑓(𝑥) corresponding to the equidistant
value of 𝑥 = 𝑥0 , 𝑥1 , 𝑥2 , … , 𝑥𝑛 , where 𝑥𝑖 − 𝑥𝑖−1 = ℎ, 𝑖 = 1, 2, … , 𝑛, then

17
 𝑢(𝑢 + 1) 2 𝑢(𝑢 + 1) … (𝑢 + (𝑛 − 1)) 𝑛
𝑦 = 𝑦𝑛 + 𝑢∇𝑦𝑛 + ∇ 𝑦𝑛 + ⋯ + ∇ 𝑦𝑛 (4)
2! 𝑛!
𝑥−𝑥𝑛
, where 𝑢 = .
ℎ

Proof:
Since the points are 𝑛 + 1, the polynomial in 𝑥 used to approximate is in degree 𝑛.
Define the polynomial as:
𝑦 = 𝑎𝑛 + 𝑎𝑛−1 (𝑥 − 𝑥𝑛 ) + 𝑎𝑛−2 (𝑥 − 𝑥𝑛 )(𝑥 − 𝑥𝑛−1 ) + ⋯
+ 𝑎0 (𝑥 − 𝑥𝑛 )(𝑥 − 𝑥𝑛−1 ) … (𝑥 − 𝑥0 ) (5)
But at 𝑥 = 𝑥𝑖 , 𝑦𝑖 = 𝑓(𝑥𝑖 ) = 𝑃𝑛 (𝑥𝑖 ).
If 𝑥 = 𝑥𝑛 , 𝑎𝑛 = 𝑦𝑛
If 𝑥 = 𝑥𝑛−1 ,
𝑦𝑛 − 𝑦𝑛−1
𝑦𝑛−1 = 𝑦𝑛 + 𝑎𝑛−1 (𝑥𝑛−1 − 𝑥𝑛 ) ⇒ 𝑦𝑛−1 − 𝑦𝑛 = 𝑎𝑛−1 (𝑥𝑛−1 − 𝑥𝑛 ) ⇒ = 𝑎𝑛−1
ℎ
∇𝑦𝑛
Therefore, 𝑎𝑛−1 = .
ℎ

If 𝑥 = 𝑥𝑛−2 ,
∇𝑦𝑛
𝑦𝑛−2 = 𝑦𝑛 + (𝑥𝑛−2 − 𝑥𝑛 ) + 𝑎2 (𝑥𝑛−2 − 𝑥𝑛 )(𝑥𝑛−1 − 𝑥𝑛 ) ⇒ 𝑦𝑛−2 = 𝑦𝑛 − 2∇𝑦𝑛 − 2ℎ2 𝑎2
ℎ
Therefore,
𝑦𝑛−2 − 𝑦𝑛 + 2∇𝑦𝑛 ∇2 𝑦𝑛
𝑎𝑛−2 = = .
2ℎ2 2! ℎ2
Similarly,
∇3 𝑦0
𝑎3 = ,
3! ℎ3
∇𝑛 𝑦0
𝑎𝑛 = .
𝑛! ℎ𝑛
Substituting all value of 𝑎𝑖 ′s in (2) we get
𝑢(𝑢 + 1) 2 𝑢(𝑢 + 1) … (𝑢 + (𝑛 − 1)) 𝑛
𝑦 = 𝑦𝑛 + 𝑢∇𝑦𝑛 + ∇ 𝑦𝑛 + ⋯ + ∇ 𝑦𝑛
2! 𝑛!
𝑛
𝑢−𝑛 𝑘
= ∑( ) ∇ 𝑦𝑛 (6)
𝑘
𝑘=0
𝑥−𝑥𝑛
, where 𝑢 = .
ℎ

Example: 𝑓(𝑥) is a function such that

18
 𝑓(0) = 6, 𝑓(1) = 5 = 𝑓(2), 𝑓(3) = 15, 𝑓(4) = 50 find 𝑓(0.5)

Solution: The difference table is as below.

𝒙 𝒇(𝒙) ∆𝒇 ∆𝟐 𝒇 ∆𝟑 𝒇 ∆𝟒 𝒇
𝟎 6
-1
𝟏 5 1
0 9
𝟐 5 10 6
10 15
𝟑 15 25
35
𝟒 50

Then using the Newton forward difference interpolating formula, we get:

𝑢(𝑢 − 1) 2 𝑢(𝑢 − 1)(𝑢 − 2) 3 𝑢(𝑢 − 1)(𝑢 − 2)(𝑢 − 3) 4

𝑦 ≈ 𝑦0 + 𝑢∆𝑦0 + ∆ 𝑦0 + ∆ 𝑦0 + ∆ 𝑦0
2! 3! 4!
𝑥−𝑥0 0.5−0
Where 𝑢 = = = 0.5.
ℎ 1

0.5(0.5) 0.5(0.5)(−1.5) 0.5(0.5)(−1.5)(−2.5)

𝑓(0.5) ≈ 0.5 + 0.5(−1) + (1) + (9) + (6)
2! 3! 4!
545
=
64

Or

∆𝑓0 ∆2 𝑓0 ∆3 𝑓0
𝑓(𝑥)𝑃4 ( 𝑥) = 𝑓0 + (𝑥 − 𝑥0 ) + (𝑥 − 𝑥0 )(𝑥 − 𝑥1 ) + (𝑥 − 𝑥0 )(𝑥 – 𝑥1 )(𝑥 – 𝑥2 )
ℎ 2! ℎ2 3! ℎ3
∆4 𝑓0
+ (𝑥 − 𝑥0 )(𝑥 – 𝑥1 )(𝑥 – 𝑥2 )(𝑥 – 𝑥3 )
4! ℎ4

1 9 6
= 6 − 1 (𝑥 − 0) + (𝑥 − 0)(𝑥 − 1) + (𝑥 − 0)(𝑥 − 1)(𝑥 – 2) + (𝑥 − 0)(𝑥 − 1)(𝑥 – 2)(𝑥 – 3)
2! 3! 4!
1
= ( 2𝑥 4 − 9𝑥 3 + 14𝑥 2 + 9𝑥 + 12)
2

19
Then,

1 1 1 1 4 1 3 1 2 1 545
𝑓 ( ) ≈ 𝑃4 ( ) = (2 ( ) − 9 ( ) + 14 ( ) + 9 ( ) + 12) =
2 2 2 2 2 2 2 64

Example: f(x) is a function such that

f(0) = 6, f(1) = 5 = f(2)

f(3) = 15, f(4) = 50

Exercises

1. The population of a country in the censing as under Estimate the population for the year 1925
Year x: 1891 1901 1911 1921 1931

Population y: 46 66 81 93 101

(in thousands)

2. Establish the relations:

 = E, = E-1

1 1
𝐸 −2  = 𝐸 2  = 

 =  =  -  = 2

1
 = 2 ( + )

1
2 = 1 + 42

3. Find the polynomial satisfied by

(-4, 1245), (-1, 33), (0, 5), (2, 9) and (5, 1335)

20
4. Show that the nth divided difference of xn is 1.

5. The function y = f(x) is given at the points (7, 3), (8, 1), (9, 1) and (10, 5). Find the value of y for x
= 9.5 using Lagrange’s interpolation formula

6. Use Gauss’s forward formulas to find the value of y when

x = 3.75 given the table

x 2.5 3.0 3.5 4.0 4.5 5.0

u -2 -1 0 1 2 3

21