Optimal Points for Interpolation 1

5.4 Optimal Points for Interpolation

1. Prove each of the following properties of the Chebyshev polynomials:

(a) for each n, Tn (1) = 1.
(b) for each n, Tn (−1) = (−1)n .
1
(c) for all j > k ≥ 0, Tj (x)Tk (x) = 2 [Tj+k (x) + Tj−k (x)].

(a) Tn (1) = cos(n cos−1 (1)) = cos 0 = 1. Alternately, we proceed by induction

on n. For n = 0, T0 (x) = 1 so that T0 (1) = 1. Similarly, T1 (x) = x so
that T1 (1) = 1. Now, suppose that for some natural number n, Tn (1) =
Tn−1 (1) = 1. Then, by the recurrence relation for the Chebyshev polynomials,

Tn+1 (1) = 2(1)Tn (1) − Tn−1 (1) = 2 − 1 = 1.

Hence, Tn (1) = 1 for all n.

(b) We proceed by induction on n. For n = 0, T0 (x) = 1 so that T0 (−1) = 1 =
(−1)0 . Similarly, T1 (x) = x so that T1 (−1) = −1 = (−1)1 . Now, suppose
that for some natural number n, Tn (−1) = (−1)n and Tn−1 (−1) = (−1)n−1 .
Then, by the recurrence relation for the Chebyshev polynomials,

Tn+1 (−1) = 2(−1)Tn (−1) − Tn−1 (−1) = 2(−1)n+1 − (−1)n−1 = (−1)n+1 .

Hence, Tn (−1) = (−1)n for all n.

(c) Here, we make use of the trigonometric identity

1
cos mx cos nx = [cos(m + n)x + cos(m − n)x].
2
Let j > k ≥ 0. Then

Tj (x)Tk (x) = cos(j cos−1 x) cos(k cos−1 x)

1
= [cos((j + k) cos−1 x) + cos((j − k) cos−1 x)]
2
1
= (Tj+k (x) + Tj−k (x)) .
2
 2 Section 5.4

2. Show that the Chebyshev polynomial Tn (x) is a solution to the differential

equation
d2 y dy
(1 − x2 ) 2 − x + n2 y = 0.
dx dx

Let y = Tn (x) = cos(n cos−1 x). Then

dy n sin(n cos−1 x)
= √ ; and
dx 1 − x2
d2 y nx sin(n cos−1 x) n2 cos(n cos−1 x)
= − .
dx2 (1 − x2 )3/2 1 − x2

Thus,
d2 y dy nx sin(n cos−1 x)
(1 − x2 ) 2
−x + n2 y = √ − n2 cos(n cos−1 x) −
dx dx 1 − x2
nx sin(n cos−1 x)
√ + n2 cos(n cos−1 x)
1 − x2
= 0,

so Tn (x) is a solution of the indicated differential equation.

3. Show that
1 
Tn (x)Tm (x)
Z
0, m 6= n
√ dx = ,
−1 1 − x2 cn π2 , m=n
where c0 = 2 and cn = 1 (n ≥ 1). This implies that the Chebyshev polynomials
form an orthogonal set on [−1, 1] with respect to the weight function w(x) =
(1 − x2 )−1/2 . (Hint: Make the substitution θ = cos−1 x.)

With the substitution θ = cos−1 x, we have

Z 1 Z π
Tn (x)Tm (x)
√ dx = cos nθ cos mθ dθ
−1 1 − x2 0
1 π
Z
= [cos(m + n)θ cos(m − n)θ] dθ.
2 0
If m 6= n, then
Z 1   π
Tn (x)Tm (x) 1 sin(m + n)θ sin(m − n)θ
√ dx = +
−1 1 − x2 2 m+n m−n 0
= 0.

On the other hand, if m = n, then

1 π
Z 1
Tn (x)Tm (x)
Z
√ dx = (1 + cos 2nθ) dθ.
−1 1 − x2 2 0
 Optimal Points for Interpolation 3

Thus, for n = 0,
1 π
Tn (x)Tm (x) 1 π
Z Z
√ dx = 2 dθ = 2 · ,
−1 1−x 2 2 0 2

while, for n ≥ 1,
Z 1   π
Tn (x)Tm (x) 1 1 π π
√ dx = θ+ sin 2nθ = =1· .
−1 1 − x2 2 2n 0 2 2

4. Show that the Legendre polynomial Pn (x) is a solution to the differential equa-
tion
d2 y dy
(1 − x2 ) 2 − 2x + n(n + 1)y = 0.
dx dx

We proceed by induction n. Observe that

d2 P0 dP0
(1 − x2 ) − 2x + 0(1)P0 = (1 − x2 )(0) − 2x(0) + 0 = 0
dx2 dx
and
d2 P1 dP1
(1 − x2 ) 2
− 2x + 1(2)P1 = (1 − x2 )(0) − 2x(1) + 2x = 0.
dx dx
Now suppose that Pn−1 (x) and Pn−2 (x) satisfy the appropriate differential equa-
tion. With
2n − 1
n−1 ′
Pn′ (x) = ′
xPn−1 (x) + Pn−1 (x) − Pn−2 (x)
n n
and
2n − 1
n − 1 ′′
Pn′′ (x) = ′′
xPn−1 ′
(x) + 2Pn−1 (x) − Pn−2 (x),
n n
it follows that
d2 Pn dPn
(1 − x2 ) − 2x + n(n + 1)Pn
dx2 dx
2n − 1 
= x((1 − x2 )Pn−1′′
(x) − 2xPn−1 ′
(x) + n(n + 1)Pn−1 (x))
n
+2(1 − x2 )Pn−1
′

(x) − 2xPn−1 (x)
n−1
(1 − x2 )Pn−2
 ′′ ′

− (x) − 2xPn−2 (x) + n(n + 1)Pn−2 (x)
n
2(2n − 1) 
(1 − x2 )Pn−1
′

= (x) + x(n − 1)Pn−1 (x)
n
2(n − 1)(2n − 1)
− Pn−2 (x)
n
 4 Section 5.4

To complete the induction step, we need the identity

(1 − x2 )Pn−1
′
(x) = (n − 1) [Pn−2 (x) − xPn−1 (x)] .

With this identity,

d2 Pn dPn
(1 − x2 ) − 2x + n(n + 1)Pn
dx2 dx
2(2n − 1)
= [(n − 1)Pn−2 (x) − x(n − 1)Pn−1 (x) + x(n − 1)Pn−1 (x)]
n
2(n − 1)(2n − 1)
− Pn−2 (x)
n
= 0,

as needed.

5. Consider interpolating f (x) = xe−x over [−1, 3] with a polynomial of degree at

most four.
(a) Interpolate at uniformly spaced points and at the scaled and translated
Legendre points. Determine the l∞ norm of the interpolation error for
both interpolating polynomials and compare with the l∞ norm associated
with the scaled and translated Chebyshev points.
(b) Interpolate at uniformly spaced points and at the scaled and translated
Chebyshev points. Determine the l2 norm of the interpolation error for
both interpolating polynomials and compare with the l2 norm associated
with the scaled and translated Legendre points.

Let f (x) = xe−x . From Example 5.12, we know that the polynomial of degree at
most four that interpolates f at the Chebyshev points, scaled and translated to the
interval [−1, 3] is

pC (x) = −0.06011x4 + 0.43376x3 − 1.11011x2 + 1.08627x + 0.01807;

from Example 5.13, the polynomial of degree at most four that interpolates f at
the Legendre points, scaled and translated to the interval [−1, 3] is

pL (x) = −0.05841x4 + 0.41820x3 − 1.07721x2 + 1.07882x + 0.00648.

The polynomial of degree at most four that interpolates f at the uniformly spaced
points x0 = −1, x1 = 0, x2 = 1, x3 = 2, and x4 = 3 is

pU (x) = −0.06018x4 + 0.43458x3 − 1.11502x2 + 1.10850x.

(a) The l∞ -norm of the interpolation error for each of the indicated interpolating
polynomials is summarized in the following table. Note that, as expected, the
l∞ -norm of the interpolation error is minimum for the Chebyshev points.
 Optimal Points for Interpolation 5

Uniform Chebyshev Legendre

0.07673 0.04610 0.09212
(b) The l2 -norm of the interpolation error for each of the indicated interpolating
polynomials is summarized in the following table. Note that, as expected, the
l2 -norm of the interpolation error is minimum for the Legendre points.
Uniform Chebyshev Legendre
0.06319 0.04411 0.03916

6. For each of the following intervals, identify the interpolating points that mini-
mize the l∞ and the l2 norm of ω for linear interpolation.
√
(a) [−1, 1] (b) [0, 3.5] (c) [−π, 0] (d) [− 2, 3] (e) [−2.5, 3.5]

For linear interpolation, two interpolating points are needed. Thus, we minimize
the l∞ -norm of ω(x) using the properly scaled and translated roots of T̃2 (x):
√ √
π 2 3π 2
cos = and cos =− ;
4 2 4 2
to minimize the l2 -norm of ω(x), we interpolate at the properly scaled and translated
roots of P̃2 (x) = x2 − 13 :
√ √
3 3
,− .
3 3
(a) Over [−1, 1], the l∞ -norm of ω(x) is minimized with the interpolating points
√ √
2 2
x0 = and x1 = − ;
2 2
the l2 -norm of ω(x) is minimized with the interpolating points
√ √
3 3
x0 = and x1 = − .
3 3
(b) Over [0, 3.5], the l∞ -norm of ω(x) is minimized with the interpolating points
√ √
2 2
x0 = 1.75 + 1.75 and x1 = 1.75 − 1.75 ;
2 2
the l2 -norm of ω(x) is minimized with the interpolating points
√ √
3 3
x0 = 1.75 + 1.75 and x1 = 1.75 − 1.75 .
3 3
(c) Over [−π, 0], the l∞ -norm of ω(x) is minimized with the interpolating points
√ √
π π 2 π π 2
x0 = − + · and x1 = − − · ;
2 2 2 2 2 2
the l2 -norm of ω(x) is minimized with the interpolating points
√ √
π π 3 π π 3
x0 = − + · and x1 = − − · .
2 2 3 2 2 3
 6 Section 5.4
√
(d) Over [− 2, 3], the l∞ -norm of ω(x) is minimized with the interpolating points
√ √ √ √ √ √
3− 2 3+ 2 2 3− 2 3+ 2 2
x0 = + · and x1 = − · ;
2 2 2 2 2 2
the l2 -norm of ω(x) is minimized with the interpolating points
√ √ √ √ √ √
3− 2 3+ 2 3 3− 2 3+ 2 3
x0 = + · and x1 = − · .
2 2 3 2 2 3

(e) Over [−2.5, 3.5], the l∞ -norm of ω(x) is minimized with the interpolating
points √ √
1 2 1 2
x0 = + 3 and x1 = − 3 ;
2 2 2 2
the l2 -norm of ω(x) is minimized with the interpolating points
√ √
1 3 1 3
x0 = + 3 and x1 = − 3 .
2 3 2 3

7. Repeat Exercise 6 for cubic interpolation.

For cubic interpolation, four interpolating points are needed. Thus, we minimize
the l∞ -norm of ω(x) using the properly scaled and translated roots of T̃4 (x):

π 3π 5π 7π
cos = 0.923880, cos = 0.382683, cos = −0.382683, cos = −0.923880;
8 8 8 8
to minimize the l2 -norm of ω(x), we interpolate at the properly scaled and translated
roots of P̃4 (x) = x4 − 76 x2 + 35
3
:

0.861136, 0.339981, −0.339981, −0.861136.

(a) Over [−1, 1], the l∞ -norm of ω(x) is minimized with the interpolating points

x0 = 0.923880, x1 = 0.382683, x2 = −0.382683, x3 = −0.923880;

the l2 -norm of ω(x) is minimized with the interpolating points

x0 = 0.861136, x1 = 0.339981, x2 = −0.339981, x3 = −0.861136.

(b) Over [0, 3.5], the l∞ -norm of ω(x) is minimized with the interpolating points

x0 = 1.75 + 1.75(0.923880) = 3.366790,

x1 = 1.75 + 1.75(0.382683) = 2.419695,
x2 = 1.75 + 1.75(−0.382683) = 1.080305,
x3 = 1.75 + 1.75(−0.923880) = 0.133210;
Optimal Points for Interpolation 7

the l2 -norm of ω(x) is minimized with the interpolating points

x0 = 1.75 + 1.75(0.861136) = 3.256988,
x1 = 1.75 + 1.75(0.339981) = 2.344967,
x2 = 1.75 + 1.75(−0.339981) = 1.122033,
x3 = 1.75 + 1.75(−0.861136) = 0.243012.
(c) Over [−π, 0], the l∞ -norm of ω(x) is minimized with the interpolating points
π π
x0 = − + (0.923880) = −0.119569,
2 2
π π
x1 = − + (0.382683) = −0.969679,
2 2
π π
x2 = − + (−0.382683) = −2.171913,
2 2
π π
x3 = − + (−0.923880) = −3.022024;
2 2
the l2 -norm of ω(x) is minimized with the interpolating points
π π
x0 = − + (0.861136) = −0.218127,
2 2
π π
x1 = − + (0.339981) = −1.036755,
2 2
π π
x2 = − + (−0.339981) = −2.104837,
2 2
π π
x3 = − + (−0.861136) = −2.923466.
2 2
√
(d) Over [− 2, 3], the l∞ -norm of ω(x) is minimized with the interpolating points
√ √
3− 2 3+ 2
x0 = + (0.923880) = 2.831995,
2√ 2√
3− 2 3+ 2
x1 = + (0.382683) = 1.637515,
2√ 2√
3− 2 3+ 2
x2 = + (−0.382683) = −0.051729,
2√ 2√
3− 2 3+ 2
x3 = + (−0.923880) = −1.246209;
2 2
the l2 -norm of ω(x) is minimized with the interpolating points
√ √
3− 2 3+ 2
x0 = + (0.861136) = 2.693512,
2√ 2√
3− 2 3+ 2
x1 = + (0.339981) = 1.543268,
2√ 2√
3− 2 3+ 2
x2 = + (−0.339981) = 0.042519,
2√ 2√
3− 2 3+ 2
x3 = + (−0.861136) = −1.107726.
2 2
 8 Section 5.4

(e) Over [−2.5, 3.5], the l∞ -norm of ω(x) is minimized with the interpolating
points
1 1
x0 = + 3(0.923880) = 3.271640, x1 = + 3(0.382683) = 1.648049,
2 2
1 1
x2 = +3(−0.382683) = −0.648049, x3 = +3(−0.923880) = −2.271640;
2 2
the l2 -norm of ω(x) is minimized with the interpolating points
1 1
x0 = + 3(0.861136) = 3.083408, x1 = + 3(0.339981) = 1.519943,
2 2
1 1
x2 = +3(−0.339981) = −0.519943, x3 = +3(−0.861136) = −2.083408.
2 2

8. Repeat Exercise 6 for interpolation by polynomials of degree at most 5.

For interpolation by polynomials of degree at most 5, six interpolating points are

needed. Thus, we minimize the l∞ -norm of ω(x) using the properly scaled and
translated roots of T̃6 (x):
π π 5π
cos = 0.965926, cos = 0.707107, cos = 0.258819,
12 4 12
7π 3π 11π
cos = −0.258819, cos = −0.707107, cos = −0.965926;
12 4 12
to minimize the l2 -norm of ω(x), we interpolate at the properly scaled and translated
15 4 5 2 5
roots of P̃4 (x) = x6 − 11 x + 11 x − 231 :
0.932470, 0.661209, 0.238619, −0.238619, −0.661209, −0.932470.
(a) Over [−1, 1], the l∞ -norm of ω(x) is minimized with the interpolating points
x0 = 0.965926, x1 = 0.707107, x2 = 0.258819,
x3 = −0.258819, x4 = −0.707107, x5 = −0.965926;
the l2 -norm of ω(x) is minimized with the interpolating points
x0 = 0.932470, x1 = 0.661209, x2 = 0.238619,
x3 = −0.238619, x4 = −0.661209, x5 = −0.932470.
(b) Over [0, 3.5], the l∞ -norm of ω(x) is minimized with the interpolating points
x0 = 1.75 + 1.75(0.965926) = 3.440371,
x1 = 1.75 + 1.75(0.707107) = 2.987437,
x2 = 1.75 + 1.75(0.258819) = 2.202933,
x3 = 1.75 + 1.75(−0.258819) = 1.297067,
x4 = 1.75 + 1.75(−0.707107) = 0.512563,
x5 = 1.75 + 1.75(−0.965926) = 0.059630;
Optimal Points for Interpolation 9

the l2 -norm of ω(x) is minimized with the interpolating points

x0 = 1.75 + 1.75(0.932470) = 3.381823,
x1 = 1.75 + 1.75(0.661209) = 2.907116,
x2 = 1.75 + 1.75(0.238619) = 2.167583,
x3 = 1.75 + 1.75(−0.238619) = 1.332417,
x4 = 1.75 + 1.75(−0.661209) = 0.592884,
x5 = 1.75 + 1.75(−0.932470) = 0.118178.

(c) Over [−π, 0], the l∞ -norm of ω(x) is minimized with the interpolating points
π π
x0 = − + (0.965926) = −0.053523,
2 2
π π
x1 = − + (0.707107) = −0.460075,
2 2
π π
x2 = − + (0.258819) = −1.164244,
2 2
π π
x3 = − + (−0.258819) = −1.977348,
2 2
π π
x4 = − + (−0.707107) = −2.681517,
2 2
π π
x5 = − + (−0.965926) = −3.088069;
2 2
the l2 -norm of ω(x) is minimized with the interpolating points
π π
x0 = − + (0.932470) = −0.106076,
2 2
π π
x1 = − + (0.661209) = −0.532172,
2 2
π π
x2 = − + (0.238619) = −1.195974,
2 2
π π
x3 = − + (−0.238619) = −1.945618,
2 2
π π
x4 = − + (−0.661209) = −2.609421,
2 2
π π
x5 = − + (−0.932470) = −3.035517.
2 2
√
(d) Over [− 2, 3], the l∞ -norm of ω(x) is minimized with the interpolating points
√ √
3− 2 3+ 2
x0 = + (0.965926) = 2.924795,
2√ 2√
3− 2 3+ 2
x1 = + (0.707107) = 2.353554,
2√ 2√
3− 2 3+ 2
x2 = + (0.258819) = 1.364134,
2√ 2√
3− 2 3+ 2
x3 = + (−0.258819) = 0.221652,
2 2
10 Section 5.4
√ √
3− 2 3+ 2
x4 = + (−0.707107) = −0.767767,
2√ 2√
3− 2 3+ 2
x5 = + (−0.965926) = −1.339009;
2 2
the l2 -norm of ω(x) is minimized with the interpolating points
√ √
3− 2 3+ 2
x0 = + (0.932470) = 2.850954,
2√ 2√
3− 2 3+ 2
x1 = + (0.661209) = 2.252252,
2√ 2√
3− 2 3+ 2
x2 = + (0.238619) = 1.319551,
2√ 2√
3− 2 3+ 2
x3 = + (−0.238619) = 0.266236,
2√ 2√
3− 2 3+ 2
x4 = + (−0.661209) = −0.666466,
2√ 2√
3− 2 3+ 2
x5 = + (−0.932470) = −1.265168.
2 2

(e) Over [−2.5, 3.5], the l∞ -norm of ω(x) is minimized with the interpolating
points
1 1
x0 = + 3(0.965926) = 3.397778, x1 = + 3(0.707107) = 2.621321,
2 2
1 1
x2 = + 3(0.258819) = 1.276457, x3 = + 3(−0.258819) = −0.276457,
2 2
1 1
x4 = +3(−0.707107) = −1.621321, x5 = +3(−0.965926) = −2.397778;
2 2
the l2 -norm of ω(x) is minimized with the interpolating points

1 1
x0 = + 3(0.932470) = 3.297410, x1 = + 3(0.661209) = 2.483627,
2 2
1 1
x2 = + 3(0.238619) = 1.215857, x3 = + 3(−0.238619) = −0.215857,
2 2
1 1
x4 = +3(−0.661209) = −1.483627, x5 = +3(−0.932470) = −2.297410.
2 2

For Exercises 9 - 13, interpolate the given function over the specified interval by
a polynomial of the indicated degree. Interpolate at uniformly spaced points,
the Chebyshev points and the Legendre points, and compare the errors in the
resulting polynomials in both the l∞ and the l2 norm.
 Optimal Points for Interpolation 11

9. f (x) = ex , [−1, 1], n = 3

Let f (x) = ex . With an interval of [−1, 1] and n = 3, the uniformly spaced

interpolating points are
1 1
x0 = −1, x1 = − , x2 = , x3 = 1
3 3
whereas the Chebyshev points are
π 3π 5π 7π
x0 = cos , x1 = cos , x2 = cos , x3 = cos
8 8 8 8
and the Legendre points are

x0 = 0.861136, x1 = 0.339981, x2 = −0.339981, x3 = −0.861136.

The corresponding interpolating polynomials are

pU (x) = 0.176152x3 + 0.547885x2 + 0.999049x + 0.995196,

pC (x) = 0.175176x3 + 0.542901x2 + 0.998933x + 0.994615,
pL (x) = 0.173940x3 + 0.536628x2 + 0.999271x + 0.996325

The table below lists the l∞ and l2 norms of the interpolation error for each of these
polynomials. As expected, the l∞ norm of the interpolation error is a minimum for
the Chebyshev points, and the l2 norm of the interpolation error is a minimum for
the Legendre points.
Chebyshev Legendre Uniform
l∞ norm 0.006657 0.012118 0.009985
l2 norm 0.005433 0.004745 0.007682

10. f (x) = e−x , [−1, 2], n = 3

Let f (x) = e−x . With an interval of [−1, 2] and n = 3, the uniformly spaced
interpolating points are

x0 = −1, x1 = 0, x2 = 1, x3 = 2

whereas the Chebyshev points are

1 3 π 1 3 3π 1 3 5π 1 3 7π
x0 = + cos , x1 = + cos , x2 = + cos , x3 = + cos
2 2 8 2 2 8 2 2 8 2 2 8
and the Legendre points are
1 3 1 3
x0 = + (0.861136), x1 = + (0.339981),
2 2 2 2
1 3 1 3
x2 = + (−0.339981), x3 = + (−0.861136).
2 2 2 2
 12 Section 5.4

The corresponding interpolating polynomials are

pU (x) = −0.114431x3 + 0.543081x2 − 1.060770x + 1.000000,

pC (x) = −0.113009x3 + 0.533503x2 − 1.051903x + 0.995997,
pL (x) = −0.111241x3 + 0.521718x2 − 1.042521x + 0.999574

The table below lists the l∞ and l2 norms of the interpolation error for each of these
polynomials. As expected, the l∞ norm of the interpolation error is a minimum for
the Chebyshev points, and the l2 norm of the interpolation error is a minimum for
the Legendre points.

Chebyshev Legendre Uniform

l∞ norm 0.023870 0.043228 0.034855
l2 norm 0.021741 0.019102 0.031129

11. f (x) = x ln x, [1, 3], n = 4

Let f (x) = x ln x. With an interval of [1, 3] and n = 4, the uniformly spaced

interpolating points are

x0 = 1, x1 = 1.5, x2 = 2, x3 = 2.5, x4 = 3

whereas the Chebyshev points are

π 3π π 7π 9π
x0 = 2+cos , x1 = 2+cos , x2 = 2+cos , x3 = 2+cos , x4 = 2+cos
10 10 2 10 10
and the Legendre points are

x0 = 2+0.906180, x1 = 2+0.538469, x2 = 2, x3 = 2−0.538469, x4 = 2−0.906180.

The corresponding interpolating polynomials are

pU (x) = 0.011937x4 − 0.141641x3 + 0.813054x2 − 0.240430x − 0.442920,

pC (x) = 0.011923x4 − 0.141500x3 + 0.812447x2 − 0.239047x − 0.444178,
pL (x) = 0.011734x4 − 0.139438x3 + 0.804711x2 − 0.227084x − 0.450627

The table below lists the l∞ and l2 norms of the interpolation error for each of these
polynomials. As expected, the l∞ norm of the interpolation error is a minimum for
the Chebyshev points, and the l2 norm of the interpolation error is a minimum for
the Legendre points.

Chebyshev Legendre Uniform

l∞ norm 0.0003537 0.0007042 0.0005868
l2 norm 0.0002461 0.0002162 0.0003428
 Optimal Points for Interpolation 13

12. f (x) = ln(x + 2), [−1, 1], n = 5

Let f (x) = ln(x + 2). With an interval of [−1, 1] and n = 5, the uniformly spaced
interpolating points are
3 1 1 3
x0 = −1, x1 = − , x2 = − , x3 = , x4 = , x5 = 1
5 5 5 5
whereas the Chebyshev points are
π π 5π 7π 3π 11π
x0 = cos , x1 = cos , x2 = cos , x3 = cos , x4 = cos , x5 = cos
12 4 12 12 4 12
and the Legendre points are

x0 = 0.932470, x1 = 0.661209, x2 = 0.238619,

x3 = −0.238619, x4 = −0.661209, x5 = −0.932470.

The corresponding interpolating polynomials are

pU (x) = 0.008238x5 − 0.020224x4 + 0.041046x3 − 0.123567x2 +

0.500022x + 0.693097,
pC (x) = 0.008386x5 − 0.020566x4 + 0.040812x3 − 0.123027x2 +
0.500048x + 0.693036,
pL (x) = 0.008149x5 − 0.020021x4 + 0.040996x3 − 0.123450x2 +
0.500032x + 0.693073

The table below lists the l∞ and l2 norms of the interpolation error for each of these
polynomials. As expected, the l∞ norm of the interpolation error is a minimum for
the Chebyshev points, and the l2 norm of the interpolation error is a minimum for
the Legendre points.
Chebyshev Legendre Uniform
l∞ norm 0.0001975 0.0004244 0.0003892
l2 norm 0.0001227 0.0001098 0.0001905

13. f (x) = 1/x, [1, 4], n = 5

Let f (x) = x−1 . With an interval of [1, 4] and n = 5, the uniformly spaced
interpolating points are
8 11 14 17
x0 = 1, x1 = , x2 = , x3 = , x4 = , x5 = 4
5 5 5 5
whereas the Chebyshev points are
5 3 π 5 3 π 5 3 5π
x0 = + cos , x1 = + cos , x2 = + cos ,
2 2 12 2 2 4 2 2 12
14 Section 5.4

5 3 7π 5 3 3π 5 3 11π
x3 =
+ cos , x4 = + cos , x5 = + cos
2 2 12 2 2 4 2 2 12
and the Legendre points are
5 3 5 3 5 3
x0 = + (0.932470), x1 = + (0.661209), x2 = + (0.238619),
2 2 2 2 2 2
5 3 5 3 5 3
x3 = + (−0.238619), x4 = + (−0.661209), x5 = + (−0.932470).
2 2 2 2 2 2
The corresponding interpolating polynomials are

pU (x) = −0.007460x5 + 0.111906x4 − 0.675910x3 + 2.096364x2 −

3.505706x + 2.980806,
pC (x) = −0.007707x5 + 0.115610x4 − 0.696550x3 + 2.148419x2 −
3.562504x + 2.999989,
pL (x) = −0.007224x5 + 0.108362x4 − 0.655096x3 + 2.035886x2 −
3.418318x + 2.930691

The table below lists the l∞ and l2 norms of the interpolation error for each of these
polynomials. As expected, the l∞ norm of the interpolation error is a minimum for
the Chebyshev points, and the l2 norm of the interpolation error is a minimum for
the Legendre points.

Chebyshev Legendre Uniform

l∞ norm 0.002743 0.005699 0.004945
l2 norm 0.001688 0.001538 0.002841