CHAPTER 4: INTERPOLATION

Hoang Hai Ha

HCMUT-OISP
Email: [email protected]

March 20, 2023

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 1 / 49

O UTLINE

1 I NTRODUCTION

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 2 / 49

O UTLINE

1 I NTRODUCTION

2 P OLYNOMIAL INTERPOLATION

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 2 / 49

O UTLINE

1 I NTRODUCTION

2 P OLYNOMIAL INTERPOLATION

3 C UBIC SPLINES

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 2 / 49

O UTLINE

1 I NTRODUCTION

2 P OLYNOMIAL INTERPOLATION

3 C UBIC SPLINES

4 L EAST SQUARE METHOD

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 2 / 49

A PROBLEM

A census of the population of the US is taken every 10 years. The

following table lists the population, in thousands of people, from
1960 to 2010 and the data are also represented in the figure

Year 1960 1970 1980 1990 2000 2010

Population 179.2 203.3 226.5 249.6 281.4 308.7

The question is: Find the population of the US in 1975.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 3 / 49

D EFINITION 1.1
The process constructing a function that fit known data points is
called interpolation and is the subject of this chapter.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 4 / 49

P OLYNOMIAL INTERPOLATION

We are given function y = f (x) by a table of data

xi x0 x1 x2 ... xn
y i = f (x i ) y0 y1 y2 ... yn

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 5 / 49

P OLYNOMIAL INTERPOLATION

We are given function y = f (x) by a table of data

xi x0 x1 x2 ... xn
y i = f (x i ) y0 y1 y2 ... yn

We will find a polynomial of degree n or smaller

P n (x) = a n x n + a n−1 x n−1 + . . . + a 1 x + a 0

that passes through all points (x i , y i ), i = 0..n, that is, satisfies

P n (x i ) = y i , i = 0, 1, 2, . . . , n

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 5 / 49

P OLYNOMIAL INTERPOLATION

We are given function y = f (x) by a table of data

xi x0 x1 x2 ... xn
y i = f (x i ) y0 y1 y2 ... yn

We will find a polynomial of degree n or smaller

P n (x) = a n x n + a n−1 x n−1 + . . . + a 1 x + a 0

that passes through all points (x i , y i ), i = 0..n, that is, satisfies

P n (x i ) = y i , i = 0, 1, 2, . . . , n

P n (x) is called the polynomial interpolation for f (x), x i is called

node.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 5 / 49

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 6 / 49
E XAMPLE 2.1
Determine the polynomial interpolation that fits the table
x 0 1 3
.
y 1 -1 2

Solution.
Polynomial has the form P (x) = a 0 x 2 + a 1 x + a 2

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 7 / 49

E XAMPLE 2.1
Determine the polynomial interpolation that fits the table
x 0 1 3
.
y 1 -1 2

Solution.
Polynomial has the form P (x) = a 0 x 2 + a 1 x + a 2

 a2 = 1
P (x i ) = y i , then a 0 + a 1 + a 2 = −1
9a 0 + 3a 1 + a 2 = 2


Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 7 / 49

E XAMPLE 2.1
Determine the polynomial interpolation that fits the table
x 0 1 3
.
y 1 -1 2

Solution.
Polynomial has the form P (x) = a 0 x 2 + a 1 x + a 2

 a2 = 1
P (x i ) = y i , then a 0 + a 1 + a 2 = −1
9a 0 + 3a 1 + a 2 = 2


T HEOREM 2.1
A function y = f (x) is presented by the table of data
x x0 x1 x2 . . . xn
y y0 y1 y2 . . . yn
Then there is a unique polynomial of maximum degree n P n (x) that
satisfies P n (x i ) = y i .
Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 7 / 49
L AGRANGE POLYNOMIAL

Given a function y = f (x) by the table

x x0 x1 x2 . . . xn
y y0 y1 y2 . . . yn
We compute the polynomial interpolation [x 0 , x n ], n Ê 1.
The Lagrange polynomial interpolation is given as follows
n
L n (x) =
X
p k (x).y k ,
k=0

where
(x − x 0 )(x − x 1 ) . . . (x − x k−1 )(x − x k+1 ) . . . (x − x n )
p k (x) = ,
(x k − x 0 )(x k − x 1 ) . . . (x k − x k−1 )(x k − x k+1 ) . . . (x k − x n )

where k = 0 . . . n − 1.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 8 / 49

E XAMPLE 2.2
x 0 1 3
Given a function y = f (x) by the table: . Evaluate
y 1 -1 2
approximate value f (2) by using Lagrange polynomial interpolation.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 9 / 49

E XAMPLE 2.2
x 0 1 3
Given a function y = f (x) by the table: . Evaluate
y 1 -1 2
approximate value f (2) by using Lagrange polynomial interpolation.

Solution
(x − 1)(x − 3) (x − 0)(x − 3) (x − 0)(x − 1)
L 2 (x) = 1× +(−1)× +2× .
(0 − 1)(0 − 3) (1 − 0)(1 − 3) (3 − 0)(3 − 1)
f (2) ≈ −2/3.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 9 / 49

A NOTHER REPRESENTATION

Let

ω(x) = (x − x 0 ) . . . (x − x k−1 )(x − x k )(x − x k+1 ) . . . (x − x n ).

Then
ω(x)
p nk (x) =
ω′ (x k )(x − x k )
The Lagrange interpolation becomes
n yk n y
k
L n (x) = ω(x). ω(x).
X X
= ,
k=0 ω′ (x )(x − x )
k k D
k=0 k

with D k = ω′ (x k )(x − x k ).

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 10 / 49

 x x0 x1 ... xn
x0 x − x0 x0 − x1 ... x0 − xn D0
x1 x1 − x0 x − x1 ... x1 − xn D1
... ... ... ... ... ...
xn xn − x0 xn − x1 ... x − xn Dn

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 11 / 49

 x x0 x1 ... xn
x0 x − x0 x0 − x1 ... x0 − xn D0
x1 x1 − x0 x − x1 ... x1 − xn D1
... ... ... ... ... ...
xn xn − x0 xn − x1 ... x − xn Dn

n
ω(x) =
Q
(x − x i ) .
i =1

D k : product of all entries on the kth row.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 11 / 49

E XAMPLE 2.3
x 0 1 3 4
Given function y = f (x) that is defined by table .
y 1 1 2 -1
Using Lagrange polynomial interpolation, approximate the value of f
at x = 2.
Solution.
2 0 1 3 4
0 2 − 0 0 − 1 0 − 3 0 − 4 D 0 = −24
1 1−0 2−1 1−3 1−4 D1 = 6
3 3−0 3−1 2−3 3−4 D2 = 6
4 4 − 0 4 − 1 4 − 3 2 − 4 D 3 = −24.
w(x) = 4.
µ ¶
1 1 1 1
f (2) ≈ 4 − × 1 + × 1 + × 2 − × (−1) .
24 6 6 24

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 12 / 49

D IVIDED DIFFERENCES

D EFINITION 2.1
The first divided difference of f respect to x k , x k+1 (denoted
y k+1 − y k
f [x k , x k+1 ]) is defined by: f [x k , x k+1 ] = .
x k+1 − x k

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 13 / 49

D IVIDED DIFFERENCES

D EFINITION 2.1
The first divided difference of f respect to x k , x k+1 (denoted
y k+1 − y k
f [x k , x k+1 ]) is defined by: f [x k , x k+1 ] = .
x k+1 − x k
The second divided difference, f [x k , x k+1 , x k+2 ] is defined :
f [x k+1 , x k+2 ] − f [x k , x k+1 ]
f [x k , x k+1 , x k+2 ] = .
x k+2 − x k

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 13 / 49

D IVIDED DIFFERENCES

D EFINITION 2.1
The first divided difference of f respect to x k , x k+1 (denoted
y k+1 − y k
f [x k , x k+1 ]) is defined by: f [x k , x k+1 ] = .
x k+1 − x k
The second divided difference, f [x k , x k+1 , x k+2 ] is defined :
f [x k+1 , x k+2 ] − f [x k , x k+1 ]
f [x k , x k+1 , x k+2 ] = .
x k+2 − x k
The p t h divided difference, f [x k , x k+1 ..., x k+p ] is defined
f [x k , x k+1 , . . . , x k+p ] =
f [x k+1 , x k+2 , . . . , x k+p ] − f [x k , x k+1 , . . . , x k+p−1 ]
.
x k+p − x k

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 13 / 49

D IVIDED DIFFERENCES

D EFINITION 2.1
The first divided difference of f respect to x k , x k+1 (denoted
y k+1 − y k
f [x k , x k+1 ]) is defined by: f [x k , x k+1 ] = .
x k+1 − x k
The second divided difference, f [x k , x k+1 , x k+2 ] is defined :
f [x k+1 , x k+2 ] − f [x k , x k+1 ]
f [x k , x k+1 , x k+2 ] = .
x k+2 − x k
The p t h divided difference, f [x k , x k+1 ..., x k+p ] is defined
f [x k , x k+1 , . . . , x k+p ] =
f [x k+1 , x k+2 , . . . , x k+p ] − f [x k , x k+1 , . . . , x k+p−1 ]
.
x k+p − x k

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 13 / 49

E XAMPLE 2.4
Find all divided differences of y = f (x) given by the table
x 1.0 1.3 1.6 1.9
y 0.76 0.62 0.45 0.28

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 14 / 49

xk yk 1st DD
1 0.76

1.3 0.62

1.6 0.45

1.9 0.28

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 15 / 49

xk yk 1st DD
1 0.76

-7/15

1.3 0.62

−17/30

1.6 0.45

-17/30

1.9 0.28

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 16 / 49

xk yk 1st DD 2nd DD 3rd DD
1 0.76

-7/15

1.3 0.62 -1/6

−17/30 5/27

1.6 0.45 0

-17/30

1.9 0.28

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 17 / 49

N EWTON POLYNOMIAL INTERPOLATION

We find a polynomial degree at most n that fits the data points in the
form:

N (x) = y 0 + a 0 (x − x 0 ) + a 1 (x − x 0 )(x − x 1 ) + a 2 (x − x 0 )(x − x 1 )(x − x 2 ) + ...

... + a n−1 (x − x 0 )(x − x 1 )...(x − x n−1 )

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 18 / 49

N EWTON POLYNOMIAL INTERPOLATION

We find a polynomial degree at most n that fits the data points in the
form:

N (x) = y 0 + a 0 (x − x 0 ) + a 1 (x − x 0 )(x − x 1 ) + a 2 (x − x 0 )(x − x 1 )(x − x 2 ) + ...

... + a n−1 (x − x 0 )(x − x 1 )...(x − x n−1 )

N (x 0 ) = y 0

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 18 / 49

N EWTON POLYNOMIAL INTERPOLATION

We find a polynomial degree at most n that fits the data points in the
form:

N (x) = y 0 + a 0 (x − x 0 ) + a 1 (x − x 0 )(x − x 1 ) + a 2 (x − x 0 )(x − x 1 )(x − x 2 ) + ...

... + a n−1 (x − x 0 )(x − x 1 )...(x − x n−1 )

N (x 0 ) = y 0
y1 − y0
N (x 1 ) = y 1 = y 0 + a 0 (x 1 − x 0 ), ⇒ a 0 = = f [x 0 , x 1 ].
x1 − x0

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 18 / 49

N EWTON POLYNOMIAL INTERPOLATION

We find a polynomial degree at most n that fits the data points in the
form:

N (x) = y 0 + a 0 (x − x 0 ) + a 1 (x − x 0 )(x − x 1 ) + a 2 (x − x 0 )(x − x 1 )(x − x 2 ) + ...

... + a n−1 (x − x 0 )(x − x 1 )...(x − x n−1 )

N (x 0 ) = y 0
y1 − y0
N (x 1 ) = y 1 = y 0 + a 0 (x 1 − x 0 ), ⇒ a 0 = = f [x 0 , x 1 ].
x1 − x0
N (x 2 ) = y 2 = y 0 + f [x 0 , x 1 ](x 2 − x 0 ) + a 1 (x 2 − x 0 )(x 2 − x 1 )

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 18 / 49

 y 2 − y 1 + y 1 − y 0 − f [x 0 , x 1 ](x 2 − x 0 )
⇒ a1 =
(x 2 − x 0 )(x 2 − x 1 )
f [x 2 , x 1 ] f [x 0 , x 1 ](x 1 − x 0 ) − f [x 0 , x 1 ](x 2 − x 0 )
= +
x2 − x0 (x 2 − x 0 )(x 2 − x 1 )
f [x 2 , x 1 ] − f [x 0 , x 1 ]
=
x2 − x0
= f [x 0 , x 1 , x 2 ].

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 19 / 49

 y 2 − y 1 + y 1 − y 0 − f [x 0 , x 1 ](x 2 − x 0 )
⇒ a1 =
(x 2 − x 0 )(x 2 − x 1 )
f [x 2 , x 1 ] f [x 0 , x 1 ](x 1 − x 0 ) − f [x 0 , x 1 ](x 2 − x 0 )
= +
x2 − x0 (x 2 − x 0 )(x 2 − x 1 )
f [x 2 , x 1 ] − f [x 0 , x 1 ]
=
x2 − x0
= f [x 0 , x 1 , x 2 ].

Similarly, we get a n−1 = f [x 0 , x 1 , ..., x n ].

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 19 / 49

F ORWARD N EWTON POLYNOMIAL INTERPOLATION

N (x)= y 0 + f [x 0 , x 1 ](x − x 0 ) + f [x 0 , x 1 , x 2 ](x − x 0 )(x − x 1 )+

+... + f [x 0 , x 1 , ...x n ](x − x 0 )(x − x 1 )...(x − x n−1 ).

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 20 / 49

F ORWARD N EWTON POLYNOMIAL INTERPOLATION

N (x)= y 0 + f [x 0 , x 1 ](x − x 0 ) + f [x 0 , x 1 , x 2 ](x − x 0 )(x − x 1 )+

+... + f [x 0 , x 1 , ...x n ](x − x 0 )(x − x 1 )...(x − x n−1 ).

B ACKWARD N EWTON POLYNOMIAL INTERPOLATION

N (x) =y n + f [x n , x n−1 ](x − x n )+

+ f [x n , x n−1 , x n−2 ](x − x n )(x − x n−1 )
+... + f [x n , x n−1 , x n−2 , ..., x 0 ](x − x n )(x − x n−1 )...(x − x 1 ).

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 20 / 49

E XAMPLE 2.5
Interpolate the following table of data points by using the Forward
Newton Interpolation:

x 1.0 1.3 1.6 1.9

y 0.76 0.62 0.45 0.28

Solution.
7 1 5
N (x) = 0.76 − (x−1)− (x−1)(x−1.3)+ (x−1)(x−1.3)(x−1.6).
15 6 17

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 21 / 49

E XAMPLE 2.6
The following data are given for an interpolating polynomial
unknown degree.

x 0 1 2
.
P (x) 2 −1 4

Determine the coefficient before x 2 if all 3r d DD are 1.

Solution.
The polynomial has form:
P (x) = 2 + f [0, 1]x + f [0, 1, 2]x(x − 1) + f [0, 1, 2, x 3 ] x(x − 1)(x − 2) .
| {z }
=1

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 22 / 49

E XAMPLE 2.6
The following data are given for an interpolating polynomial
unknown degree.

x 0 1 2
.
P (x) 2 −1 4

Determine the coefficient before x 2 if all 3r d DD are 1.

Solution.
The polynomial has form:
P (x) = 2 + f [0, 1]x + f [0, 1, 2]x(x − 1) + f [0, 1, 2, x 3 ] x(x − 1)(x − 2) .
| {z }
=1
2
The coefficient before x is : f [0, 1, 2] − 3 f [0, 1, 2, x 3 ] .

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 22 / 49

E RROR

T HEOREM 2.2
Suppose that x 0 , x 1 , . . . x n are distinct points and P (x) is the
interpolating polynomial of a differentiable function of order n + 2
f (x), then we have the estimate
¯ ¯
¯ (n+1) (x − x 0 )(x − x 1 ) . . . (x − x n ) ¯
| f (x) − P (x)| ≤ max ¯ f ¯ (x) ¯.
[a,b] (n + 1)! ¯

E XAMPLE 2.7
Use 3 nodes x = 2, x = 2.75, x = 4 on [2, 4], estimate the maximum
1
error of the interpolating polynomial P (x) of the f (x) = .
x

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 23 / 49

Solution of example 2.7
¯ ¯
¯ (3) (x − 2)(x − 2.75)(x − 4) ¯
The maximum error is max ¯ f (x)
¯ ¯.
[2,4] 3! ¯

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 24 / 49

Solution of example 2.7
¯ ¯
¯ (3) (x − 2)(x − 2.75)(x − 4) ¯
The maximum error is max ¯ f (x)
¯ ¯.
[2,4] 3! ¯
3
max | f (3) (x)| = max |6x −4 | = .
[2,4] [2,4] 8
g (x) = (x − 2)(x − 2.75)(x − 4) = x 3 − (35x 2 )/4 + (49)/2 − 22, we
have max |g (x)| = |g (7/2)|.
[2,4]
1
Thus, the maximum error is · |g (7/2)|
16

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 24 / 49

W HY DO WE USE CUBIC SPLINES ?
Disadvantage of Lagrange and Newton polynomial is that
high-degree polynomials can oscillate erratically( a minor
fluctuation over a small portion of the interval can induce large
fluctuations over the entire range).

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 25 / 49

C UBIC SPLINE INTERPOLATION

An alternative approach is to divide the approximation interval into

a collection of subintervals and construct a (generally) different
approximating polynomial on each subinterval. This is called
piecewise-polynomial approximation.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 26 / 49

C UBIC SPLINE INTERPOLATION

An alternative approach is to divide the approximation interval into

a collection of subintervals and construct a (generally) different
approximating polynomial on each subinterval. This is called
piecewise-polynomial approximation.
D EFINITION 3.1
The piecewise-polynomial approximation uses polynomials degree 3
between each successive pair of nodes is called cubic spline
interpolation.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 26 / 49

T HE CASE OF 3 NODES
x x0 x1 x2
Given a function y = f (x) defined by table
f (x) y0 y1 y2
We construct the piecewise polynomial as:
(
S 0 (x) = a 0 + b 0 (x − x 0 ) + c 0 (x − x 0 )2 + d 0 (x − x 0 )3 , x ∈ [x 0 , x 1 ]
S(x) = ,
S 1 (x) = a 1 + b 1 (x − x 1 ) + c 1 (x − x 1 )2 + d 1 (x − x 1 )3 , x ∈ [x 1 , x 2 ]

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 27 / 49

T HE CASE OF 3 NODES
x x0 x1 x2
Given a function y = f (x) defined by table
f (x) y0 y1 y2
We construct the piecewise polynomial as:
(
S 0 (x) = a 0 + b 0 (x − x 0 ) + c 0 (x − x 0 )2 + d 0 (x − x 0 )3 , x ∈ [x 0 , x 1 ]
S(x) = ,
S 1 (x) = a 1 + b 1 (x − x 1 ) + c 1 (x − x 1 )2 + d 1 (x − x 1 )3 , x ∈ [x 1 , x 2 ]

and satisfying the following conditions:

S(x i ) = y i .

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 27 / 49

T HE CASE OF 3 NODES
x x0 x1 x2
Given a function y = f (x) defined by table
f (x) y0 y1 y2
We construct the piecewise polynomial as:
(
S 0 (x) = a 0 + b 0 (x − x 0 ) + c 0 (x − x 0 )2 + d 0 (x − x 0 )3 , x ∈ [x 0 , x 1 ]
S(x) = ,
S 1 (x) = a 1 + b 1 (x − x 1 ) + c 1 (x − x 1 )2 + d 1 (x − x 1 )3 , x ∈ [x 1 , x 2 ]

and satisfying the following conditions:

S(x i ) = y i .
S(x) has continuous second derivative on [x 0 , x 2 ].
S 0 (x 0 ) = y 0 , S 0 (x 1 ) = y 1 , S 1 (x 1 ) = y 1 , S 2 (x 2 ) = y 2 .
The first derivative is continuous : S 0′ (x 1 ) = S 1′ (x 1 )
The second derivative is continuous: S 0′′ (x 1 ) = S 1′′ (x 1 )
⇒ 6 equations with 8 variables ⇒ we need to implement 2 equations.
The piecewise function found as above is called cubic spline
interpolation.
Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 27 / 49
Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 28 / 49
G ENERAL CASE

Given function y = f (x) by data points in the table:

x x 0 x 1 ... x n
f (x) y 0 y 1 ... y n

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 29 / 49

G ENERAL CASE

Given function y = f (x) by data points in the table:

x x 0 x 1 ... x n
f (x) y 0 y 1 ... y n
Constructing cubic spline of the data set above is to find:

S 0 (x),


x ∈ [x 0 , x 1 ]
S 1 (x),

x ∈ [x 1 , x 2 ]
S(x) = .

..


S n−1 (x), x ∈ [x n−1 , x n ]


where S i (x) = a i + b i (x − x i ) + c i (x − x i )2 + d i (x − x i )3 , i = 0...n − 1, ⇒

4n unknowns, 4n − 2 equations.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 29 / 49

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 30 / 49
N ATURAL CUBIC SPLINE

N ATURAL CUBIC SPLINE : S ′′ (x 0 ) = 0, S ′′ (x n ) = 0

Let h i = x i +1 − x i , i = 0..n − 1.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 31 / 49

N ATURAL CUBIC SPLINE

N ATURAL CUBIC SPLINE : S ′′ (x 0 ) = 0, S ′′ (x n ) = 0

Let h i = x i +1 − x i , i = 0..n − 1.
S ′′ (x 0 ) = 0, ⇒ c 0 = 0

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 31 / 49

N ATURAL CUBIC SPLINE

N ATURAL CUBIC SPLINE : S ′′ (x 0 ) = 0, S ′′ (x n ) = 0

Let h i = x i +1 − x i , i = 0..n − 1.
S ′′ (x 0 ) = 0, ⇒ c 0 = 0
Define c n = S ′′ (x n )/2, ⇒ c n = 0

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 31 / 49

Find vector C = (c 0 , c 1 , c 2 , . . . , c n−1 , c n )T = (0, c 1 , c 2 , . . . , 0)T from the
system AC = B , where

1 0 0 ... 0 0
 

 h 0 2(h 0 + h 1 ) h 1 . . . 0 0 

A=
 ... ... ... ... ... ... 

 0 0 0 . . . 2(h n−2 + h n−1 ) h n−1 
0 0 0 ... 0 1

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 32 / 49

Find vector C = (c 0 , c 1 , c 2 , . . . , c n−1 , c n )T = (0, c 1 , c 2 , . . . , 0)T from the
system AC = B , where

1 0 0 ... 0 0
 

 h 0 2(h 0 + h 1 ) h 1 . . . 0 0 

A=
 ... ... ... ... ... ... 

 0 0 0 . . . 2(h n−2 + h n−1 ) h n−1 
0 0 0 ... 0 1
 
0
y2 − y1 y1 − y0
3 −3
 
h1 h0
 
 
B = ...
 

 y n − y n−1 y n−1 − y n−2 
 3 −3 
 h n−1 h n−2

0

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 32 / 49

After finding vector C , remaining coefficients a k , b k , d k are defined
by

ak = yk


y k+1 − y k h k



bk = − (c k+1 + 2c k )
hk 3

 c k+1 − c k
 dk
 = , ∀k = 0..n − 1
3h k

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 33 / 49

E XAMPLE 3.1
x 0 2 5
Construct natural cubic spline from the table
y 1 1 4

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 34 / 49

E XAMPLE 3.1
x 0 2 5
Construct natural cubic spline from the table
y 1 1 4
     
1 0 0 0 0
Solution: A = 2 10 3 , B = 3 , ⇒ c = 0.3
    
0 0 1 0 0
a 0 = 1, a 1 = 1

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 34 / 49

E XAMPLE 3.1
x 0 2 5
Construct natural cubic spline from the table
y 1 1 4
     
1 0 0 0 0
Solution: A = 2 10 3 , B = 3 , ⇒ c = 0.3
    
0 0 1 0 0
a 0 = 1, a 1 = 1
b 0 = −0.2, b 1 = 0.4

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 34 / 49

E XAMPLE 3.1
x 0 2 5
Construct natural cubic spline from the table
y 1 1 4
     
1 0 0 0 0
Solution: A = 2 10 3 , B = 3 , ⇒ c = 0.3
    
0 0 1 0 0
a 0 = 1, a 1 = 1
b 0 = −0.2, b 1 = 0.4
d 0 = 0.05, d 1 = −1/30

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 34 / 49

E XAMPLE 3.2
x 0 1 3 4
Given function y = f (x) by the table below . Using
y 1 2 4 8
natural cubic spline, find approximately f (1.5) and f ′ (3.5).

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 35 / 49

E XAMPLE 3.2
x 0 1 3 4
Given function y = f (x) by the table below . Using
y 1 2 4 8
natural cubic spline, find approximately f (1.5) and f ′ (3.5).
0
     
1 0 0 0 0
1 6 2 0  9
, B = 0, ⇒ c = − 16 
 
Solution A = 
0 2 6 1 9  27 
16
0 0 0 1 0 0

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 35 / 49

E XAMPLE 3.2
x 0 1 3 4
Given function y = f (x) by the table below . Using
y 1 2 4 8
natural cubic spline, find approximately f (1.5) and f ′ (3.5).
0
     
1 0 0 0 0
1 6 2 0  9
, B = 0, ⇒ c = − 16 
 
Solution A = 
0 2 6 1 9  27 
16
0 0 0 1 0 0
a 0 = 1, a 1 = 2, a 3 = 4

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 35 / 49

E XAMPLE 3.2
x 0 1 3 4
Given function y = f (x) by the table below . Using
y 1 2 4 8
natural cubic spline, find approximately f (1.5) and f ′ (3.5).
0
     
1 0 0 0 0
1 6 2 0  9
, B = 0, ⇒ c = − 16 
 
Solution A = 
0 2 6 1 9  27 
16
0 0 0 1 0 0
a 0 = 1, a 1 = 2, a 3 = 4
b 0 = 19/16, b 1 = 5/8, b 2 = 23/8

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 35 / 49

E XAMPLE 3.2
x 0 1 3 4
Given function y = f (x) by the table below . Using
y 1 2 4 8
natural cubic spline, find approximately f (1.5) and f ′ (3.5).
0
     
1 0 0 0 0
1 6 2 0  9
, B = 0, ⇒ c = − 16 
 
Solution A = 
0 2 6 1 9  27 
16
0 0 0 1 0 0
a 0 = 1, a 1 = 2, a 3 = 4
b 0 = 19/16, b 1 = 5/8, b 2 = 23/8
d 0 = −3/16, d 1 = 3/8, d 2 = −9/16

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 35 / 49

E XAMPLE 3.2
x 0 1 3 4
Given function y = f (x) by the table below . Using
y 1 2 4 8
natural cubic spline, find approximately f (1.5) and f ′ (3.5).
0
     
1 0 0 0 0
1 6 2 0  9
, B = 0, ⇒ c = − 16 
 
Solution A = 
0 2 6 1 9  27 
16
0 0 0 1 0 0
a 0 = 1, a 1 = 2, a 3 = 4
b 0 = 19/16, b 1 = 5/8, b 2 = 23/8
d 0 = −3/16, d 1 = 3/8, d 2 = −9/16
f (1.5) = 71/32, f ′ (3.5) = 265/64

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 35 / 49

C LAMPED CUBIC SPLINE
f ′ (a) = S ′ (x 0 ) = α, f ′ (b) = S ′ (x n ) = β

We define
 vector C = (c 1 , c 2 , ..., c n )T from AC = B , where 
2h 0 h0 0 0 ... 0 0

 h0 2(h 0 + h 1 ) h1 0 ... 0 0 

 0 h1 2(h 1 + h 2 ) h2 ... 0 
A=
 
... ... ... ... ... ... ...

 
 
 0 0 0 ... . . . 2(h n−2 + h n−1 ) h n−1 
0 0 0 0 ... h n−1 2h n−1 n×n
 y1 − y0 
3 − 3α
h
y 2 − y 10 y1 − y0
 
 
 3 −3 

 h1 h0 

B = ... 
y n − y n−1 y n−1 − y n−2 
 

 3 −3 
 h n−1 h n−2 
 y n − y n−1 
3β − 3
h n−1
Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 36 / 49
E XAMPLE 3.3
x 0 1 2
Construct clamped cubic spline interpolated by table
y 1 2 1
satisfying conditions f ′ (0) = 0, f ′ (2) = 0.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 37 / 49

E XAMPLE 3.3
x 0 1 2
Construct clamped cubic spline interpolated by table
y 1 2 1
satisfying conditions f ′ (0) = 0, f ′ (2) = 0.

Solution
 
2 1 0
Find c = (c 0 , c 1 , c 2 )T from Ac = B , where A = 1 4 1,
0 1 2
 
3
B = −6

3

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 37 / 49

E XAMPLE 3.3
x 0 1 2
Construct clamped cubic spline interpolated by table
y 1 2 1
satisfying conditions f ′ (0) = 0, f ′ (2) = 0.

Solution
 
2 1 0
Find c = (c 0 , c 1 , c 2 )T from Ac = B , where A = 1 4 1,
0 1 2
   
3 3
B = −6 ⇒ c = −3
  
3 3

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 37 / 49

E XAMPLE 3.3
x 0 1 2
Construct clamped cubic spline interpolated by table
y 1 2 1
satisfying conditions f ′ (0) = 0, f ′ (2) = 0.

Solution
 
2 1 0
Find c = (c 0 , c 1 , c 2 )T from Ac = B , where A = 1 4 1,
0 1 2
   
3 3
B = −6 ⇒ c = −3
  
3 3
b 0 = 0, b 1 = 0

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 37 / 49

E XAMPLE 3.3
x 0 1 2
Construct clamped cubic spline interpolated by table
y 1 2 1
satisfying conditions f ′ (0) = 0, f ′ (2) = 0.

Solution
 
2 1 0
Find c = (c 0 , c 1 , c 2 )T from Ac = B , where A = 1 4 1,
0 1 2
   
3 3
B = −6 ⇒ c = −3
  
3 3
b 0 = 0, b 1 = 0
d 0 = −2, d 0 = 2

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 37 / 49

E XAMPLE 3.4
A car traveling along a straight road is clocked at a number of points.
The data from the observations are given in the following table, where
the time is in hours, the distance is in km, and the speed is in km/h.
time 0 3 5
distance 0 100 180 . Use a clamped cubic spline to predict the
speed 75 77 80
position of the car when t = 4(h).

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 38 / 49

E XAMPLE 3.4
A car traveling along a straight road is clocked at a number of points.
The data from the observations are given in the following table, where
the time is in hours, the distance is in km, and the speed is in km/h.
time 0 3 5
distance 0 100 180 . Use a clamped cubic spline to predict the
speed 75 77 80
position of the car when t = 4(h).

SOLUTION
We construct the clamped cubic spline S(t ) to describe the
position of car at t time. From the table we have
S ′ (0) = 75, S ′ (5) = 80.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 38 / 49

S OLUTION

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 39 / 49

S OLUTION

   
6 3 0 −125
A = 3 10 2, B =  20 ,
0 2 4 120

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 39 / 49

S OLUTION

     
6 3 0 −125 −67/3
A = 3 10 2, B =  20 ,⇒ c =  3 
0 2 4 120 57/2

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 39 / 49

S OLUTION

     
6 3 0 −125 −67/3
A = 3 10 2, B =  20 ,⇒ c =  3 
0 2 4 120 57/2
b 0 = 75, b 1 = 17

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 39 / 49

S OLUTION

     
6 3 0 −125 −67/3
A = 3 10 2, B =  20 ,⇒ c =  3 
0 2 4 120 57/2
b 0 = 75, b 1 = 17
d 0 = 76/27, d 1 = 17/4

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 39 / 49

S OLUTION

     
6 3 0 −125 −67/3
A = 3 10 2, B =  20 ,⇒ c =  3 
0 2 4 120 57/2
b 0 = 75, b 1 = 17
d 0 = 76/27, d 1 = 17/4
Position when t = 4(h) is 124.25

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 39 / 49

I NTERPOLATION IN PRACTICAL PROBLEMS

On the plane Ox y, we have many large of points M k (x k , y k ),

k = 1, 2, ..., n, (at least two points are different x i ̸= x j , i ̸= j . In that
case, the construction a curve going through all these point is not
efficient.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 40 / 49

I NTERPOLATION IN PRACTICAL PROBLEMS

On the plane Ox y, we have many large of points M k (x k , y k ),

k = 1, 2, ..., n, (at least two points are different x i ̸= x j , i ̸= j . In that
case, the construction a curve going through all these point is not
efficient.
We will find an approximate function y = f (x) that describes data in
the table, but it’s not necessary for it to go exactly through given
points.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 40 / 49

L EAST SQUARE METHOD

Find minimum of the functional

n
( f (x k ) − y k )2 → mi n
X
g(f ) =
k=1

The forms of functions that we usually use are

f (x) = A + B x, f (x) = A + B x +C x 2 , f (x) = Ap(x) + B q(x)
E RROR
We use the next formula to estimate the error of the least square
method
n
( f (x k ) − y k )2 .
P
E=
i =1

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 41 / 49

F ORM f (x) = A + B x

Then
n
(A + B x k − y k )2
X
g (A, B ) =
k=1

It leads to finding extreme values of two variables function, so we

need to find critical point (solve the system of partial derivative)

∂ P n n

(A + B x k − y k )2 = 2
P

 (A + B x k − y k ) = 0
∂A k=1

k=1
∂ Pn n
(A + B x k − y k )2 = 2
P
(A + B x k − y k )x k = 0


∂B k=1

k=1

n n
 µ ¶
P P


 nA + xk B = yk
k=1 ¶ k=1
⇔ µ
n
¶ µ
n n
x k2 B =
P P P
xk A + xk y k



k=1 k=1 k=1

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 42 / 49

E XAMPLE 4.1
Find approximate function f (x) = A + B x for the given table by the
x 1 1 2 2 2
least square method and estimate the error E .
y 1 2 2 3 4

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 43 / 49

E XAMPLE 4.1
Find approximate function f (x) = A + B x for the given table by the
x 1 1 2 2 2
least square method and estimate the error E .
y 1 2 2 3 4

Solution
y = 1.5x.

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 43 / 49

E XAMPLE 4.1
Find approximate function f (x) = A + B x for the given table by the
x 1 1 2 2 2
least square method and estimate the error E .
y 1 2 2 3 4

Solution
y = 1.5x.
5
E = (y i − 1.5x i )2 .
P
i =1

E XAMPLE 4.2
Find approximate function f (x) = A ln x + B for the given table by the
least square method and linear model.
x 1 1.3 2 2 2.3
y 1.3 2 2 3.3 4.3

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 43 / 49

F ORM f (x) = A + B x +C x 2

n
(A + B x k +C x k2 − y k )2
X
g (A, B,C ) =
k=1

We find extreme value of 3 variable function

∂ n

(A + B x k +C x k2 − y k ) = 0
P
 g (A, B,C ) = 2
∂A


k=1


 ∂

n
(A + B x k +C x k2 − y k )x k = 0
P
g (A, B,C ) = 2
 ∂B k=1
∂


 n
(A + B x k +C x k2 − y k )x k2 = 0
P
g (A, B,C ) = 2


∂C

k=1

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 44 / 49

 n n n
 µ ¶ µ ¶
P P 2 P


 n A + x k B + x k
C = yk
k=1 ¶ k=1 ¶ k=1


n n n n
 µ
 P ¶ µ µ
x k2 B + x k3 C =
P P P
⇔ xk A + xk y k
 µk=1 ¶
 µk=1 ¶ µk=1 ¶ k=1

 n n n n
x2 A + x3 B + x4 C = x2 y
 P P P P
k k

k k k

k=1 k=1 k=1 k=1

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 45 / 49

E XAMPLE 4.3
Find approximate function f (x) = A + B x +C x 2 for the below table by
using least square method
x 1 1 2 3 3
y 4.12 4.18 6.23 8.34 8.38

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 46 / 49

E XAMPLE 4.3
Find approximate function f (x) = A + B x +C x 2 for the below table by
using least square method
x 1 1 2 3 3
y 4.12 4.18 6.23 8.34 8.38

E XAMPLE 4.4
Find approximate function f (x) = A sin x + B cos 2x +C for the below
table by using least square method
x 1 1.2 2.4 3.5 3.7
y 4.2 4.8 6.3 8.4 8.8

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 46 / 49

F ORM f (x) = Ap(x) + B q(x)

n
(Ap(x k ) + B q(x k ) − y k )2
X
g (A, B ) =
k=1

We need to define the extreme value of 2 variable function g (A, B )

n

∂ P
 ∂A g (A, B ) = 2 (Ap(x k ) + B q(x k ) − y k )p(x k ) = 0


k=1
n
∂ P
∂B g (A, B ) = 2 (Ap(x k ) + B q(x k ) − y k )q(x k ) = 0



k=1

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 47 / 49

  µ n ¶ µ
n
¶
n
P 2 P P


 p (x k ) A + p(x k )q(x k ) B = p(x k )y k
⇔ µk=1 ¶ k=1 µ k=1
n n n
¶
q 2 (x k ) B =
P P P
p(x k )q(x k ) A + q(x k )y k



k=1 k=1 k=1

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 48 / 49

E XAMPLE 4.5
p
Find approximate function f (x) = A x + B cos(x) for given table by
least square method
x 1.0 1.2 1.4 1.6 1.8
y 2.27 2.37 2.45 2.52 2.60

Hoang Hai Ha (BK TPHCM) CHAPTER 4: INTERPOLATION March 20, 2023 49 / 49