Chapter 3

Interpolation and Polynomial Approximation

3.1 Interpolation and the Lagrange Polynomial

3.1.1 Lagrangian Form

Consider a polynomial of degree (n  1):

P(x) = a1x n-1 + a2x n-2 +... + an-1x + an

where the ai are constants. The polynomial can be written in Lagrangian form:

P(x) = c1(x  2) (x  3)... (x  n) + c2(x  1) (x  3)... (x  n) + ...
ci(x  1) (x  2) ... (x  i-1) (x  i+1) ... (x  n) + ...
cn(x  1) (x  2)... (x  n-1)

where i, i = 1, 2, ..., n are arbitrary scalars, while the constants ci are related to the constants
ai.

Example 3.1-1 _____________________________________________________

Write the polynomial P(x) = x 2  4x + 3 in the Lagrangian form.

Solution

The Lagrangian form for P(x) = x 2  4x + 3 is

P(x) = c1(x  2) (x  3) + c2(x  1) (x  3) + c3(x  1) (x  2)

where i, i = 1, 2, 3 are arbitrary scalars. Let 1 = 1, 2 = 2, 3 = 3, then

P(x) = c1(x  2) (x  3) + c2(x  1) (x  3) + c3(x  1) (x  2)

The constants c1 can be evaluated from the above relation by substituting x = 1 = 1

P(x = 1) = 1  4 + 3 = c1(1  2) (1  3)  c1 = 0

For x = 2 = 2

P(x = 2) = 4  8 + 3 = c2(2  1) (2  3)  c2 = 1

For x = 3 = 3

1
 P(x = 3) = 9  12 + 3 = c3(3  1) (3  2)  c3 = 0

The Lagrangian form for the polynomial is

P(x) = (x  1)(x  3)

Let 1 =  2, 2 =  1, 3 = 2, then

P(x) = c1(x + 1) (x  2) + c2(x + 2) (x  2) + c3(x + 2) (x + 1)

The constants ci can be evaluated to obtain: c1 = 3.7500, c2 = -2.6667, and c3 = -0.0833. The
Lagrangian form for the polynomial is

P(x) = 3.7500 (x + 1) (x  2)  2.6667 (x + 2) (x  2)  0.0833 (x + 2) (x + 1)

A short form notation for P(x) is

P(x) =

where denotes product of all terms (x  k), for k varying from 1 to n except i.

Let x = i then

P(i) = ci(i  1) (i  2) ... (i  i-1) (i  i+1) ... (i  n)

The constant ci can be expressed as

ci =

3.1.2 Polynomial Approximation

Consider a function f(x) that passes through the two distinct points (x0, f(x0)) and (x1, f(x1)) as
shown in Figure 3.1-1. The first order polynomial that approximates the function between
these two points can be expressed as

P(x) = a + bx

where a and b are constants. P(x) can also be written in Lagrangian form as

P(x) = c0(x  x1) + c1(x  x0)

2
 Figure 3.1-1 First and second order polynomial approximation.

where

ci =

or
c0 = = , and c1 = =

The approximating polynomial is finally

P(x) = f(x0) + f(x1)

The first order polynomial basis function L0(x) is defined as

L0(x) = =

Similarly, the first order polynomial basis function L1(x) is defined as

L1(x) = =

In terms of the basis function, P(x) can be written as

P(x) = L0(x) f(x0) + L1(x) f(x1)

If a second order polynomial is used to approximate the function using three points (x0, f(x0)),
(x1, f(x1)), and (x2, f(x2)) then

3
 P(x) = f(x0) + f(x1) + f(x2)

P(x) can also be written in terms of the second order polynomial basis function L2,k(x)

P(x) = L2,0(x) f(x0) + L2,1(x)f(x1) + L2,2(x)f(x2)

where L2,0(x) = =

In general: L2,k(xk) = 1 at node k, L2,k(xi) = 0 at other nodes.

We now seek a polynomial P(x) of degree n that interpolates a given function f(x) between
the node xi of the grid for which there are n+1 nodes x0, x1, , xn and

P(xk) = f(xk) for each k = 1, 2, , n

The polynomial is given by

P(x) = Ln,0(x) f(x0) + Ln,1(x) f(x1) +  + Ln,n(x)f(xn) = f(xk)

where Ln,k(x) = ; Ln,k(xi) = 0 and Ln,k(xk) = 1

Polynomial approximation constitutes the foundation upon which we shall build the various
numerical methods. The approximation P(x) to f(x) is known as a Lagrange interpolation
polynomial, and the function Ln,k(x) is called a Lagrange basis polynomial.

Example 3.1-2 _____________________________________________________

Find the Lagrange interpolation polynomial that takes the values prescribed below

xk 0 1 2 4
f(xk) 1 1 2 5

Solution
P(x) = f(xk)

P(x) = (1) + (1)

+ (2) + (5)

4
When working with grids having large numbers of intervals one typically assigns a set of low
degree (n = 1, 2, or 3) basis functions to each adjacent set of n+1 = 2, 3, or 4 nodes.

Example 3.1-3 _____________________________________________________

Use global interpolation by one polynomial and piecewise polynomial interpolation with
quadratic for the following nodes.

xk 0 1 2 4 5
f(xk) 0 16 48 88 0

Solution

Global interpolation by one polynomial: P(x) = f(xk)

P(x) = (0) + (16)

+ (48) + (88) + 0

Piecewise polynomial interpolation with quadratic

P(x) = (0) + (16) + (48); 0x2

P(x) = (48) + (88) + (0); 2x5

The error En(x) associated with the interpolation of f(x) by Pn(x) over the interval [x0, xn] can
be estimated as

En(x) = f(x)  Pn(x) = ( )

where  is some number lying in the open interval (x0, xn) and

Wn(x) = (x  x0)(x  x1)  (x  xn)

When the spacial increments are uniform

xk+1  xk = h, k = 0, 1, 2, , n-1

Let x = x0 + h, since

5
 x1 = x0 + h  x  x1 = (  1)h

xn = x0 + nh  x  xn = (  n)h

Wn(x) = (x  x0)(x  x1)  (x  xn) = (h)[(  1)h] [(  n)h]

The error associated with interpolation is then

En(x) = () = (h)[(  1)h] [(  n)h] ( )

The only variable in the above expression is h the spacing of the nodes, therefore

En(x) = Chn+1, x0 <  < xn

where C is a coefficient independent of h.

We can therefore write En(x) = O(hn+1) meaning that the ratio En(x)/ hn+1 is bounded by a
constant as h  0. As the increment h decreases, so also will the interpolation error En.

Example 3.1-4 _____________________________________________________

For the function f(x) = ln(x + 1), construct interpolation polynomials of degree one and two to
approximate f(0.45) from the given nodes. Find the error bound and the actual error.

xk 0 0.6 0.9
ln(x + 1) 1 0.47000 0.64185

Solution

First degree polynomial

P1(x) = (0) + (0.47) = 0.78334x

P1(0.45) = 0.3525

Error bound: En(x) = (x  x0)(x  x1)  (x  xn) ()

E1(x) = | (x  x0)(x  x1)|

f(x) = ln(x + 1)  f’(x) =  f”(x) =  f””(x) =

6
 E1(x) = | (0.45  0)(0.45  0.6)| = 3.37510-2

Actual error = |ln(1 + 0.45)  P1(0.45)| = 1.90610-2

Second degree polynomial

P2(x) = (0) + (0.47)

+ (0.64185)

P2(0.45) = 0.36829

Error bound: E2(x) = | (x  x0)(x  x1)(x  x2)|

E2(x) = | (0.45  0)(0.45  0.6)(0.45  0.9)| = 1.012510-2

Actual error = |ln(1 + 0.45)  P2(0.45)| = 3.272910-3