Interpolation Example

By: Ahmad Sirojuddin, S.T., M.Sc., Ph.D.

sirojuddin@its.ac.id

Given a function as
� = � � = �0.4� − 0.4�2 (exponential plus quadratic).
Find a polynomial function that interpolates � � at � =− 1, � = 1, � = 4, � = 6, and � = 8 by using
A. Newton’s Divided-Difference Interpolating Polynomial
B. Lagrange Interpolating Polynomial
C. Linear Splines
D. Cubic Splines (with natural spline condition)

Evaluate the function at � = 3

Solution:

From the problem, we have the following data:

�0 =− 1 � �0 = 0.27032
�1 = 1 � �1 = 1.0918247
�2 = 4 � �2 =− 1.4469676
�3 = 6 � �3 =− 3.3768236
�4 = 8 � �4 =− 1.0674698

A. Newton’s Divided-Difference Interpolating Polynomial

Since we have 5 data points, then the polynomial order is 4 (� = 4). Accordingly, the polynomial form is
expressed as:
� � = �0 + �1 � − �0 + �2 � − �0 � − �1 + �3 � − �0 � − �1 � − �2
+ �4 � − �0 � − �1 � − �2 � − �3
Here, our task is to determine the values of �0 until �4. Before we calculate them, let’s calculate the finite
divided differences corresponding to the data points.
First finite divided difference equations:
� �1 − � �0 1.0918247 − 0.27032
� �1, �0 = = = 0.4107523
�1 − �0 1 − −1
� �2 − � �1 −1.4469676 − 1.0918247
� �2, �1 = = =− 0.8462641
�2 − �1 4−1
 � �3 − � �2 −3.3768236 − −1.4469676
� �3, �2 = = =− 0.964928
�3 − �2 6−4
� �4 − � �3 −1.0674698 − −3.3768236
� �4, �3 = = = 1.1546769
�4 − �3 8−6
Second finite divided difference equations:
� �2 , �1 − � �1 , �0 −0.8462641 − 0.4107523
� �2, �1 , �0 = = =− 0.2514033
�2 − �0 4 − −1
� �3 , �2 − � �2 , �1 −0.964928 − −0.8462641
� �3, �2 , �1 = = =− 0.0237328
�3 − �1 6−1
� �4 , �3 − � �3 , �2 1.1546769 − −0.964928
� �4, �3 , �2 = = = 0.5299012
�4 − �2 8−4
Third finite divided difference equations:
� �3 , �2 , �1 − � �2 , �1 , �0 −0.0237328 − −6.2570164
� �3, �2 , �1 , �0 = = = 0.0325244
�3 − �0 6 − −1
� �4 , �3 , �2 − � �3 , �2 , �1 0.5299012 − −0.0237328
� �4, �3 , �2 , �1 = = = 0.0790906
�4 − �1 8−1
Fourth finite divided difference equations:
� �4 , �3 , �2 , �1 − � �3 , �2 , �1 , �0 0.0790906 − 0.8904691
� �4, �3 , �2 , �1, �0 = = = 0.005174
�4 − �0 8 − −1

Now, we can compute the values of �0 until �4 .

�0 = � �0 = 0.27032
�1 = � �1 , �0 = 0.4107523
�2 = � �2 , �1 , �0 =− 0.2514033
�3 = � �3 , �2 , �1 , �0 = 0.0325244
�4 = � �4 , �3 , �2 , �1 , �0 = 0.005174

Therefore, the polynomial function that fit data given in the problem is stated as
� � = 0.27032 + 0.4107523 � − �0 − 0.2514033 � − �0 � − �1
+0.0325244 � − �0 � − �1 � − �2
+0.005174 � − �0 � − �1 � − �2 � − �3
� � = 0.27032 + 0.4107523 � + 1 − 0.2514033 � + 1 � − 1
+0.0325244 � + 1 � − 1 � − 4
+0.005174 � + 1 � − 1 � − 4 � − 6

The figure above shows the plot of original function � � (black curve) and the interpolating polynomial
function � � (red curve) found by using the Newton’s divided-difference method. The conclusion of the
figure:
 � � and � � intersect at � =− 1, � = 1, � = 4, � = 6, and � = 8.
 Within interval −1 ≤ � ≤ 8, the curve of � � and � � is near.
 Within interval � <− 1 or � > 8, we cannot guarantee that � � and � � is near.

The value of the function at � = 3

� � = 0.27032 + 0.4107523 3 + 1 − 0.2514033 3 + 1 3 − 1
+0.0325244 3 + 1 3 − 1 3 − 4
+0.005174 3 + 1 3 − 1 3 − 4 3 − 6
=− 0.23389

B. Lagrange Interpolating Polynomial

By using Lagrange Interpolating Polynomial, the polynomial function that fit with 5 data points (order 4) is
expressed as
� − �1 � − �2 � − �3 � − �4 � − �0 � − �2 � − �3 � − �4
� � = � �0 + � �1
�0 − �1 �0 − �2 �0 − �3 �0 − �4 �1 − �0 �1 − �2 �1 − �3 �1 − �4
� − �0 � − �1 � − �3 � − �4 � − �0 � − �1 � − �2 � − �4
+ � �2 + � �3
�2 − �0 �2 − �1 �2 − �3 �2 − �4 �3 − �0 �3 − �1 �3 − �2 �3 − �4
� − �0 � − �1 � − �2 � − �3
+ � �4
�4 − �0 �4 − �1 �4 − �2 �4 − �3
�−1 �−4 �−6 �−8 �+1 �−4 �−6 �−8
= 0.27032 + 1.0918247
−1 − 1 −1 − 4 −1 − 6 −1 − 8 1+1 1−4 1−6 1−8
�+1 �−1 �−6 �−8 �+1 �−1 �−4 �−8
+ −1.4469676 + −3.3768236
4+1 4−1 4−6 4−8 6+1 6−1 6−4 6−8
�+1 �−1 �−4 �−6
+ −1.0674698
8+1 8−1 8−4 8−6
� � = 0.0004290793 � − 1 � − 4 � − 6 � − 8 − 0.005199165 � + 1 � − 4 � − 6 � − 8
−0.01205806 � + 1 � − 1 � − 6 � − 8 + 0.02412016 � + 1 � − 1 � − 4 � − 8
−0.002117996 � + 1 � − 1 � − 4 � − 6

The figure above shows the plot of original function � � (black curve) and the interpolating polynomial
function � � (red curve) found by using the Lagrange method. As seen, � � obtained by using the Lagrange
method is the same as � � obtained by using the Newton’s Divided-Difference method (see the previous
figure).

The value of the function at � = 3

� � = 0.0004290793 3 − 1 3 − 4 3 − 6 3 − 8 − 0.005199165 3 + 1 3 − 4 3 − 6 3 − 8
−0.01205806 3 + 1 3 − 1 3 − 6 3 − 8 + 0.02412016 3 + 1 3 − 1 3 − 4 3 − 8
−0.002117996 3 + 1 3 − 1 3 − 4 3 − 6
=− 0.2339152

C. Linear Splines
Interpolating using Splines means we need to create a function for each interval of data points. Since the
problem gives us 5 data points, then there are 4 four intervals with different equations.
The general equation form for interval i by using linear splines is
� �� − � ��−1
�� � = � ��−1 + � − ��−1
�� − ��−1

Interval 1 �0 ≤ � ≤ �1 , � = 1:
�0 =− 1 � �0 = 0.27032
�1 = 1 � �1 = 1.0918247
� �1 − � �0 1.0918247 − 0.27032
� � = � �0 + � − �0 = 0.27032 + � − −1
�1 − �0 1 − −1
= 0.27032 + 0.4107524 � + 1 ; −1≤�≤1

Interval 2 �1 ≤ � ≤ �2 , � = 2:
�1 = 1 � �1 = 1.0918247
�2 = 4 � �2 =− 1.4469676
� �2 − � �1 −1.4469676 − 1.0918247
� � = � �1 + � − �1 = 1.0918247 + �−1
�2 − �1 4−1
= 1.0918247 − 0.8462641 � − 1 ; 1≤�≤4

Interval 3 �2 ≤ � ≤ �3 , � = 3:
�2 = 4 � �2 =− 1.4469676
�3 = 6 � �3 =− 3.3768236
� �3 − � �2 −3.3768236 − −1.4469676
� � = � �2 + � − �2 =− 1.4469676 + �−4
�3 − �2 6−4
=− 1.4469676 − 0.964928 � − 4 ; 4≤�≤6

Interval 4 �3 ≤ � ≤ �4 , � = 4:
�3 = 6 � �3 =− 3.3768236
�4 = 8 � �4 =− 1.0674698
� �4 − � �3 −1.0674698 − −3.3768236
� � = � �3 + � − �3 =− 3.3768236 + �−6
�4 − �3 8−6
=− 3.3768236 + 1.1546769 � − 6 ; 6≤�≤8
The value of the function at � = 3 can be obtained by using interval 2 1 ≤ � ≤ 4
� � = 1.0918247 − 0.8462641 3 − 1 =− 0.6007035

D. Cubic Splines
By using cubic splines, the equation at each interval i is given by:
�'' ��−1 3
�'' �� 3
� ��−1 �'' ��−1 �� − ��−1
�� � = �� − � + � − ��−1 + − �� − �
6 �� − ��−1 6 �� − ��−1 �� − ��−1 6
� �� �'' �� �� − ��−1
+ − � − ��−1
�� − ��−1 6
Note that the values of ��−1 , �� , � ��−1 , � �� are given by the problem. This equation contains only two
unknowns, �'' ��−1 and �'' �� , which can be found by evaluating the following equation:
�� − ��−1 �'' ��−1 + 2 ��+1 − ��−1 �'' �� + ��+1 − �� �'' ��+1
6 6
= � ��+1 − � �� + � ��−1 − � ��
��+1 − �� �� − ��−1

Let’s re-write the data given by the problem:

�0 =− 1 � �0 = 0.27032
�1 = 1 � �1 = 1.0918247
�2 = 4 � �2 =− 1.4469676
�3 = 6 � �3 =− 3.3768236
�4 = 8 � �4 =− 1.0674698

 From the case � = 1:

�1 − �0 �'' �0 + 2 �2 − �0 �'' �1 + �2 − �1 �'' �2
6 6
= � �2 − � �1 + � �0 − � �1
�2 − �1 �1 − �0
Then,
1 −− 1 �'' �0 + 2 4 −− 1 �'' �1 + 4 − 1 �'' �2
6 6
= −1.4469676 − 1.0918247 + 0.27032 − 1.0918247
4−1 1 −− 1
Then,
 2�'' �0 + 10�'' �1 + 3�'' �2 =− 7.5420987 eq. (1)

 From the case � = 2:

�2 − �1 �'' �1 + 2 �3 − �1 �'' �2 + �3 − �2 �'' �3
6 6
= � �3 − � �2 + � �1 − � �2
�3 − �2 �2 − �1
Then,
4 − 1 �'' �1 + 2 6 − 1 �'' �2 + 6 − 4 �'' �3
6 6
= −3.3768236 −− 1.4469676 + 1.0918247 −− 1.4469676
6−4 4−1
Then,
3�'' �1 + 10�'' �2 + 2�'' �3 =− 0.7119834 eq. (2)

 From the case � = 3:

�3 − �2 �'' �2 + 2 �4 − �2 �'' �3 + �4 − �3 �'' �4
6 6
= � �4 − � �3 + � �2 − � �3
�4 − �3 �3 − �2
Then,
6 − 4 �'' �2 + 2 8 − 4 �'' �3 + 8 − 6 �'' �4
6 6
= −1.0674698 −− 3.3768236 + −1.4469676 −− 3.3768236
8−6 6−4
Then,
2�'' �2 + 8�'' �3 + 2�'' �4 = 12.7176294 eq. (3)
 Because of the natural spline condition, the second derivatives of the first and last points are zero, i.e.,
�'' �0 = 0 eq. (4)
�'' �4 = 0 eq. (5)

Equation 1 − 5 form a system of linear equation, where the variables are �'' �0 until �'' �4 . Let’s put
them together in one place.
�'' �0 = 0
2�'' �0 + 10�'' �1 + 3�'' �2 =− 7.5420987
3�'' �1 + 10�'' �2 + 2�'' �3 =− 0.7119834
2�'' �2 + 8�'' �3 + 2�'' �4 = 12.7176294
�'' �4 = 0

We can use any methods to find the solution, such as Gaussian elimination and substitution, Gauss-Jordan
Elimination, etc. The details are omitted here. The solution to above system of linear equations is:
�'' �0 = 0, �'' �1 =− 0.69739262, �'' �2 =− 0.18939083, �'' �3 = 1.63705138, �'' �4 = 0

Substituting this result to �� � yields the interpolation functions as follows:

Interval 1 �0 ≤ � ≤ �1 , � = 1:
�'' �0 �'' �1 � �0 �'' �0 �1 − �0
�1 � = �1 − � 3 + � − �0 3 + − �1 − �
6 �1 − �0 6 �1 − �0 �1 − �0 6
� �1 �'' �1 �1 − �0
+ − � − �0
�1 − �0 6
0 −0.69739262 0.27032 0 1 −− 1
= 1−� 3+ � −− 1 3 + − 1−�
6 1 −− 1 6 1 −− 1 1 −− 1 6
1.0918247 −0.69739262 1 −− 1
+ − � −− 1
1 −− 1 6
=− 0.05811605 � + 1 3 + 0.13516 1 − � + 0.7783766 � + 1
 3
=− 0.05811605 � + 1 + 0.6432165� + 0.9135366

Interval 2 �1 ≤ � ≤ �2 , � = 2:
�'' �1 �'' �2 � �1 �'' �1 �2 − �1
�2 � = �2 − � 3 + � − �1 3 + − �2 − �
6 �2 − �1 6 �2 − �1 �2 − �1 6
� �2 �'' �2 �2 − �1
+ − � − �1
�2 − �1 6
−0.69739262 −0.18939083
= 4−� 3+ �−1 3
6 4−1 6 4−1
1.0918247 −0.69739262 4 − 1
+ − 4−�
4−1 6
−1.4469676 −0.18939083 4 − 1
+ − �−1
4−1 6
=− 0.03874403 4 − � 3 − 0.01052171 � − 1 3 + 0.7126379 4 − � − 0.3876271 � − 1

Interval 3 �2 ≤ � ≤ �3 , � = 3:
�'' �2 3
�'' �3 3
� �2 �'' �2 �3 − �2
�3 � = �3 − � + � − �2 + − �3 − �
6 �3 − �2 6 �3 − �2 �3 − �2 6
� �3 �'' �3 �3 − �2
+ − � − �2
�3 − �2 6
−0.18939083 1.63705138 −1.4469676 −0.18939083 6 − 4
= 6−� 3+ �−4 3+ − 6−�
6 6−4 6 6−4 6−4 6
−3.3768236 1.63705138 6 − 4
+ − �−4
6−4 6
=− 0.01578257 6 − � 3 + 0.1364209 � − 4 3 − 0.6603535 6 − � − 2.2340956 � − 4

Interval 4 �3 ≤ � ≤ �4 , � = 4:
�'' �3 3
�'' �4 3
� �3 �'' �3 �4 − �3
�4 � = �4 − � + � − �3 + − �4 − �
6 �4 − �3 6 �4 − �3 �4 − �3 6
� �4 �'' �4 �4 − �3
+ − � − �3
�4 − �3 6
1.63705138 0 −3.3768236 1.63705138 8 − 6
= 8−� 3+ �−6 3+ − 8−�
6 8−6 6 8−6 8−6 6
−1.0674698 0 8 − 6
+ − �−6
8−6 6
= 0.1364209 8 − � 3 − 2.2340956 8 − � − 0.5337349 � − 6
The value of the function at � = 3 can be obtained by using interval 2 1 ≤ � ≤ 4
�2 3 =− 0.03874403 4 − 3 3 − 0.01052171 3 − 1 3 + 0.7126379 4 − 3 − 0.3876271 3 − 1
=− 0.185534