3.4. Finite Difference Tables#

3.4.1. Forward Difference Table#

Table 3.1 A forward difference table#

\(x\)

\(f(x)\)

\(\Delta f\)

\(\Delta^2 f\)

\(\Delta^3 f\)

\(\Delta^4 f\)

\(x_{-2}\)

\(f_{-2}\)

\(\Delta f_{-2}\)

\(x_{-1}\)

\(f_{-1}\)

\(\Delta^2 f_{-2}\)

\(\Delta f_{-1}\)

\(\Delta^3 f_{-2}\)

\(x_0\)

\(f_0\)

\(\Delta^2 f_{-1}\)

\(\Delta^4 f_{-2}\)

\(\Delta f_0\)

\(\Delta^3 f_{-1}\)

\(x_1\)

\(f_1\)

\(\Delta^2 f_0\)

\(\Delta f_1\)

\(x_2\)

\(f_2\)

where \( \Delta f_0=f_1-f_0 \); \( \Delta^2 f_0=\Delta f_1-\Delta f_0 \); With this notation the same suffixes appear along diagonals running from top left to bottom right.

3.4.2. Backward Difference Table#

Table 3.2 A backward difference table#

\(x\)

\(f(x)\)

\(\nabla f\)

\(\nabla^2 f\)

\(\nabla^3 f\)

\(\nabla^4 f\)

\(x_{-2}\)

\(f_{-2}\)

\(\nabla f_{-1}\)

\(x_{-1}\)

\(f_{-1}\)

\(\nabla^2 f_{0}\)

\(\nabla f_{0}\)

\(\nabla^3 f_{ 1}\)

\(x_0\)

\(f_0\)

\(\nabla^2 f_{ 1}\)

\(\nabla^4 f_{2}\)

\(\nabla f_1\)

\(\nabla^3 f_{2}\)

\(x_1\)

\(f_1\)

\(\nabla^2 f_2\)

\(\nabla f_2\)

\(x_2\)

\(f_2\)

where \( \nabla f_0 = f_0 - f_{-1} \); \( \nabla^2 f_0 = \nabla f_0 - \nabla f_{-1} \); Note that with this notation the same suffixes appear along diagonals running from bottom left to top right of the difference table. Note that \( \Delta f_1 =\nabla f_2 \) and \( \Delta^2 f_0 =\nabla^2 f_2 .\)

Remark 3.2

At same locations in a difference table, the forward and backward differences are the same. For example, if the sequences \(\{f_n\} (n\in \Z)\) given in Table 3.1 and Table 3.2 are the same, then

\[\begin{split} \Delta f_{-2} & = \nabla f_{-1} \\ \nabla f_2 & = \Delta f_1\\ \Delta^2 f_{-1} & = \nabla^2 f_{ 1} \\ \Delta^3 f_{-2} & = \nabla^3 f_{ 1} \\ \Delta^4 f_{-2} & = \nabla^4 f_{2}\\ & \vdots \\ \end{split}\]

3.4.3. Use Difference Table for interpolation#

Example 3.3

Given the following sequence \(\left\{f_j\right\}_{j=0}^{4}\) with \(h=1\), find \(f(0.1)\) and \(f(3.8)\).

\(j\)

\(x_j\)

\(f_j=f(x_j)\)

\(1^\text{st}\)diff

\(2^\text{nd}\)diff

\(3^\text{rd}\)diff

\(4^\text{th}\)diff

0

0

1

……

1

1

2

……

……

……

2

2

4

……

……

……

……

3

3

8

……

……

4

4

16

Example 3.4

From the tabulated values of the function \(f(x)\) given below interpolate a value for \(f(0.55)\).

\(x\)

\(f(x)\)

\(\Delta f\)

\(\Delta^2 f\)

\(\Delta^3 f\)

\(\Delta^4 f\)

\(\Delta^5 f\)

\(\Delta^6 f\)

0.5

0.47943

0.16479

0.7

0.64422

-0.02568

0.13911

-0.00555

0.9

0.78333

-0.03123

0.00124

0.10788

-0.00430

0.00017

1.1

0.89121

-0.03553

0.00142

-0.00006

0.07235

-0.00288

0.00011

1.3

0.96356

-0.03841

0.00153

0.03394

-0.00135

1.5

0.99749

-0.03976

-0.00583

1.7

0.99166

3.4.4. Difference Table and Polynomial Function#

\(x\)

\(f(x)\)

\(\Delta f\)

\(\Delta^2 f\)

\(\Delta^3 f\)

\(\Delta^4 f\)

0

0

1

1

1

6

7

6

2

8

12

0

19

6

3

27

18

0

37

6

4

64

24

0

61

6

5

125

30

91

6

216

  • \(3^\text{rd}\) order differences are constant;

  • the function can be approximated by a \(3^\text{rd}\) order polynomial \(f(x)=x^3\);

  • In order to find a polynomial of degree \(n\) which approximates the tabulated function sufficiently accurate, we need to know the function values at \((n+1)\) points.