AMA 3201: Computational Methods

Subject Lecturer: Jianbo Cui

Lecture 6
Numerical integration
Composite rules

0 / 25
 Problem Setting

Rb
Aim: Given f ∈ C[a, b], compute a
f (x)dx.

1 / 25
 Problem Setting

Rb
Aim: Given f ∈ C[a, b], compute a
f (x)dx.
Rb P
Idea: Construct xi ∈ [a, b] and αi so that a
f (x)dx ≈ αi f (xi ).

1 / 25
 Problem Setting

Rb
Aim: Given f ∈ C[a, b], compute a
f (x)dx.
Rb P
Idea: Construct xi ∈ [a, b] and αi so that a
f (x)dx ≈ αi f (xi ).

Two approaches:
• Equally spaced xi , based on interpolating polynomials.
Examples: rectangular rule, trapezoidal rule, Simpson’s ( 13 ) rule.
• Unequally spaced xi , based on osculating polynomials.
Example: Gaussian quadrature.

1 / 25
 Rectangular Rule

Idea:
• Partition [a, b] into

a =: x0 < x1 < · · · < xn−1 < xn := b

with equally spaced xi .

2 / 25
 Rectangular Rule

Idea:
• Partition [a, b] into

a =: x0 < x1 < · · · < xn−1 < xn := b

with equally spaced xi .

• On each [xi , xi+1 ], approximate f by the constant f ( xi +x2 i+1 ).

2 / 25
 Rectangular Rule

Idea:
• Partition [a, b] into

a =: x0 < x1 < · · · < xn−1 < xn := b

with equally spaced xi .

• On each [xi , xi+1 ], approximate f by the constant f ( xi +x2 i+1 ).
• (Composite rule) Sum the formula up for i = 0, . . . , n − 1.

2 / 25
 Rectangular Rule

b−a
Formula: On each [xi , xi+1 ], write h = n = xi+1 − xi . We have
Z xi+1    
xi + xi+1 xi + xi+1
f dx = f h.
xi 2 2

3 / 25
 Rectangular Rule

b−a
Formula: On each [xi , xi+1 ], write h = n = xi+1 − xi . We have
Z xi+1    
xi + xi+1 xi + xi+1
f dx = f h.
xi 2 2

Thus,
Z b n−1 Z xi+1   n−1  
X xi + xi+1 X xi + xi+1
f (x)dx ≈ f dx = f h.
a xi 2 2
i=0 i=0

3 / 25
 Rectangular Rule

Truncation error: Assume that f ∈ C 2 [a, b].

xi +xi+1
On each [xi , xi+1 ], write ui := 2 . Then we have
Z xi+1 Z xi+1
[f (x) − f (ui )]dx = f [ui , x] (x − ui ) dx.
xi xi | {z }
changes sign

4 / 25
 Rectangular Rule

Truncation error: Assume that f ∈ C 2 [a, b].

xi +xi+1
On each [xi , xi+1 ], write ui := 2 . Then we have
Z xi+1 Z xi+1
[f (x) − f (ui )]dx = f [ui , x] (x − ui ) dx.
xi xi | {z }
changes sign

Rx
Define r (x) := xi
(y − ui )dy. Then r (xi ) = 0 and
xi+1
1
r (xi+1 ) = (y − ui )2 = 0.
2 xi

Moreover, r ′ (x) is negative on (xi , ui ) and positive on (ui , xi+1 ) =⇒

max[xi ,xi+1 ] r (x) ≤ max{r (xi ), r (xi+1 )} = 0.

4 / 25
 Rectangular Rule

Truncation error cont.: Hence, we have

Z xi+1 Z xi+1
f [ui , x](x − ui )dx = f [ui , x]r ′ (x)dx
xi xi
xi+1 Z xi+1
= r (x)f [ui , x] − f [ui , x, x]r (x)dx
xi xi
Z xi+1
= −f [ui , ξi , ξi ] r (x)dx for some ξi ∈ [xi , xi+1 ],
xi
Z xi+1
1
= − f [ui , ξi , ξi ] [(x − ui )2 − (xi − ui )2 ]dx
2 xi
h3 h3
=− f [ui , ξi , ξi ] = − f ′′ (µi ) for some µi ∈ [xi , xi+1 ].
12 24

5 / 25
 Rectangular Rule

Truncation error cont.: Total truncation error is

n−1 Z
X xi+1
f [ui , x](x − ui )dx
i=0 xi

n−1 3 n−1
X h ′′ h3 1 X ′′
= f (µi ) = ·n· f (µi )
24 24 n
i=0 i=0
(b − a)h2 ′′
= |f (µ)|
24
for some µ ∈ [a, b].

6 / 25
 Example
Rπ
Example: Compute 02 sin x dx by the rectangular rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 0.011.

7 / 25
 Example
Rπ
Example: Compute 02 sin x dx by the rectangular rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 0.011.

iπ π
Solution: Note that f (x) = sin x, xi = 8 for i = 0, . . . , 4, and h = 8.

7 / 25
 Example
Rπ
Example: Compute 02 sin x dx by the rectangular rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 0.011.

iπ π
Solution: Note that f (x) = sin x, xi = 8 for i = 0, . . . , 4, and h = 8.
Z π         
2 π π 3π 5π 7π
sin x dx ≈ sin + sin + sin + sin
0 8 16 16 16 16
= 1.0065.

7 / 25
 Example
Rπ
Example: Compute 02 sin x dx by the rectangular rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 0.011.

iπ π
Solution: Note that f (x) = sin x, xi = 8 for i = 0, . . . , 4, and h = 8.
Z π         
2 π π 3π 5π 7π
sin x dx ≈ sin + sin + sin + sin
0 8 16 16 16 16
= 1.0065.

(b − a)h2 ′′
Error = |f (µ)| for some µ ∈ [0, π2 ],
24
π π 2
( )
≤ 2 8 ≈ 0.0101 < 0.011.
24

7 / 25
 Trapezoidal Rule

Idea:
• Partition [a, b] into

a =: x0 < x1 < · · · < xn−1 < xn := b

with equally spaced xi .

8 / 25
 Trapezoidal Rule

Idea:
• Partition [a, b] into

a =: x0 < x1 < · · · < xn−1 < xn := b

with equally spaced xi .

• On each [xi , xi+1 ], approximate f by the interpolating polynomial
p1 at xi and xi+1 .

8 / 25
 Trapezoidal Rule

Idea:
• Partition [a, b] into

a =: x0 < x1 < · · · < xn−1 < xn := b

with equally spaced xi .

• On each [xi , xi+1 ], approximate f by the interpolating polynomial
p1 at xi and xi+1 .
• (Composite rule) Sum the formula up for i = 0, . . . , n − 1.

8 / 25
 Trapezoidal Rule
Formula: On each [xi , xi+1 ], write h = b−a n = xi+1 − xi . We have
Z xi+1 Z xi+1  
x − xi x − xi+1
p1 (x) dx = fi+1 + fi dx.
xi xi xi+1 − xi xi − xi+1
Z xi+1  
x − xi x − xi+1
= fi+1 − fi dx
xi h h
1   xi+1
= (x − xi )2 fi+1 − (x − xi+1 )2 fi
2h xi
   
1 2 fi+1 + f i fi+1 + fi
= (xi+1 − xi ) = h.
h 2 2

9 / 25
 Trapezoidal Rule
Formula: On each [xi , xi+1 ], write h = b−a n = xi+1 − xi . We have
Z xi+1 Z xi+1  
x − xi x − xi+1
p1 (x) dx = fi+1 + fi dx.
xi xi xi+1 − xi xi − xi+1
Z xi+1  
x − xi x − xi+1
= fi+1 − fi dx
xi h h
1   xi+1
= (x − xi )2 fi+1 − (x − xi+1 )2 fi
2h xi
   
1 2 fi+1 + f i fi+1 + fi
= (xi+1 − xi ) = h.
h 2 2
Thus,
Z b n−1    
X fi+1 + fi f0 + fn
f (x)dx ≈ h= + f1 + f2 + · · · + fn−1 h.
a 2 2
i=0

9 / 25
 Trapezoidal Rule
Truncation error: Assume that f ∈ C 2 [a, b].
On each [xi , xi+1 ], we have
Z xi+1 Z xi+1
[f (x) − p1 (x)]dx = f [xi , xi+1 , x] (x − xi )(x − xi+1 ) dx
xi xi | {z }
≤0
Z xi+1
= f [xi , xi+1 , ξi ] (x − xi )(x − xi+1 )dx for some ξi ∈ [xi , xi+1 ],
xi
Z xi+1
= f [xi , xi+1 , ξi ] (x − xi )(x − xi − h)dx
xi
Z xi+1
= f [xi , xi+1 , ξi ] [(x − xi )2 − (x − xi )h]dx
xi
h3 h3
=− f [xi , xi+1 , ξi ] = − f ′′ (νi ) for some νi ∈ [xi , xi+1 ].
6 12

10 / 25
 Trapezoidal Rule
Truncation error cont.: Total truncation error is
n−1 Z
X xi+1
f [xi , xi+1 , x](x − xi )(x − xi+1 )dx
i=0 xi

n−1 3 n−1
X h ′′ h3 1 X ′′
= f (νi ) = ·n· f (νi )
12 12 n
i=0 i=0
2
(b − a)h ′′
= |f (ν)|
12
for some ν ∈ [a, b].

11 / 25
 Trapezoidal Rule
Truncation error cont.: Total truncation error is
n−1 Z
X xi+1
f [xi , xi+1 , x](x − xi )(x − xi+1 )dx
i=0 xi

n−1 3 n−1
X h ′′ h3 1 X ′′
= f (νi ) = ·n· f (νi )
12 12 n
i=0 i=0
2
(b − a)h ′′
= |f (ν)|
12
for some ν ∈ [a, b].

Remark: Note that this error seems larger than that of rectangular
rule, which is counter-intuitive. However, note that the ν here and the
µ for the rectangular rule can be radically different: given that both
errors are in the order of h2 , this can have a big impact on the error.

11 / 25
 Example
Rπ
Example: Compute 02 sin x dx by the trapezoidal rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 0.021.

12 / 25
 Example
Rπ
Example: Compute 02 sin x dx by the trapezoidal rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 0.021.

iπ π
Solution: Note that f (x) = sin x, xi = 8 for i = 0, . . . , 4, and h = 8.

12 / 25
 Example
Rπ
Example: Compute 02 sin x dx by the trapezoidal rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 0.021.

iπ π
Solution: Note that f (x) = sin x, xi = 8 for i = 0, . . . , 4, and h = 8.
π
π sin 0 + sin π2
Z  
2 π π 3π
sin x dx ≈ + sin + sin + sin
0 8 2 8 4 8
= 0.9871.

12 / 25
 Example
Rπ
Example: Compute 02 sin x dx by the trapezoidal rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 0.021.

iπ π
Solution: Note that f (x) = sin x, xi = 8 for i = 0, . . . , 4, and h = 8.
π
π sin 0 + sin π2
Z  
2 π π 3π
sin x dx ≈ + sin + sin + sin
0 8 2 8 4 8
= 0.9871.

(b − a)h2 ′′
Error = |f (ν)| for some ν ∈ [0, π2 ]
12
π π 2
( )
≤ 2 8 ≈ 0.0202 < 0.021.
12

12 / 25
 Simpson’s Rule

Idea:
• Partition [a, b] into

a =: x0 < x1 < · · · < xn−1 < xn := b

with equally spaced xi . Note that n has to be even!

13 / 25
 Simpson’s Rule

Idea:
• Partition [a, b] into

a =: x0 < x1 < · · · < xn−1 < xn := b

with equally spaced xi . Note that n has to be even!

• On each [xi , xi+2 ], i even, approximate f by the interpolating
polynomial p2 at xi , xi+1 and xi+2 .

13 / 25
 Simpson’s Rule

Idea:
• Partition [a, b] into

a =: x0 < x1 < · · · < xn−1 < xn := b

with equally spaced xi . Note that n has to be even!

• On each [xi , xi+2 ], i even, approximate f by the interpolating
polynomial p2 at xi , xi+1 and xi+2 .
• (Composite rule) Sum the formula up for i = 0, 2, 4, . . . , n − 2.

13 / 25
 Simpson’s Rule

b−a
Formula: On each [xi , xi+2 ], i even, write h = n = xi+1 − xi .
Consider
(x − xi+1 )(x − xi+2 ) (x − xi )(x − xi+2 )
p2 (x) = fi + fi+1
(xi − xi+1 )(xi − xi+2 ) (xi+1 − xi )(xi+1 − xi+2 )
(x − xi )(x − xi+1 )
+ fi+2
(xi+2 − xi )(xi+2 − xi+1 )
(x − xi+1 )(x − xi+2 ) (x − xi )(x − xi+2 )
= fi + fi+1
2h2 −h2
(x − xi )(x − xi+1 )
+ fi+2 .
2h2
We integrate the three quadratics one by one on [xi , xi+2 ].

14 / 25
 Simpson’s Rule

Formula cont.: Let x = xi+1 + sh for s ∈ [−1, 1]. Then we have

xi+2 Z 1
(x − xi+1 )(x − xi+2 ) s(s − 1)
Z
h
2
dx = h ds = ,
xi 2h −1 2 3
xi+2 Z 1
(x − xi )(x − xi+2 ) (s + 1)(s − 1)
Z
4h
dx = h ds = ,
xi −h2 −1 −1 3
xi+2 Z 1
(x − xi )(x − xi+1 )
Z
(s + 1)s h
dx = h ds = ,
xi 2h2 −1 2 3

Hence,
Z xi+2
h
p2 (x) dx = (fi + 4fi+1 + fi+2 ).
xi 3

15 / 25
 Simpson’s Rule

Formula cont.: Thus, the integral is given approximately by

Z b
f (x) dx
a
h
≈ [(f0 + fn ) + 2(f2 + f4 + f6 + · · · + fn−2 ) + 4(f1 + f3 + · · · + fn−1 )].
3

16 / 25
 Simpson’s Rule
Truncation error: Assume that f ∈ C 4 [a, b].

On each [xi , xi+2 ], i even, we have

Z xi+2 Z xi+2
[f (x)−p2 (x)]dx = f [xi , xi+1 , xi+2 , x] (x − xi )(x − xi+1 )(x − xi+2 ) dx.
xi xi | {z }
changes sign

17 / 25
 Simpson’s Rule
Truncation error: Assume that f ∈ C 4 [a, b].

On each [xi , xi+2 ], i even, we have

Z xi+2 Z xi+2
[f (x)−p2 (x)]dx = f [xi , xi+1 , xi+2 , x] (x − xi )(x − xi+1 )(x − xi+2 ) dx.
xi xi | {z }
changes sign

Rx
Define r (x) := xi (y − xi )(y − xi+1 )(y − xi+2 )dy . Then r (xi ) = 0 and
(substituting y = xi+1 + sh)
Z 1
r (xi+2 ) = h4 (s + 1)s(s − 1)ds = 0.
−1

Moreover, r ′ (x) is positive on (xi , xi+1 ) and negative on (xi+1 , xi+2 )

=⇒ min[xi ,xi+2 ] r (x) ≥ min{r (xi ), r (xi+2 )} = 0.

17 / 25
 Simpson’s Rule

Truncation error cont.: Hence, we have

Z xi+2
f [xi , xi+1 , xi+2 , x](x − xi )(x − xi+1 )(x − xi+2 )dx
xi
Z xi+2
= f [xi , xi+1 , xi+2 , x]r ′ (x)dx
xi
xi+2 Z xi+1
= r (x)f [xi , xi+1 , xi+2 , x] − f [xi , xi+1 , xi+2 , x, x]r (x)dx
xi xi
Z xi+2
= −f [xi , xi+1 , xi+2 , ζi , ζi ] r (x)dx for some ζi ∈ [xi , xi+2 ].
xi

18 / 25
 Simpson’s Rule

Truncation error cont.: Next, observe that

Z xi+2
r (x)dx
xi
xi+2 Z xi+2
= (x − xi+1 )r (x) − r ′ (x)(x − xi+1 )dx
xi xi
Z xi+2
=− (x − xi )(x − xi+1 )2 (x − xi+2 )dx substitute x = xi+1 + sh
xi
Z 1 Z 1
= −h5 (s + 1)s2 (s − 1)ds = −h5 (s4 − s2 )ds
−1 −1
4h5
= .
15

19 / 25
 Simpson’s Rule
Truncation error cont.: Consequently,
Z xi+2
f [xi , xi+1 , xi+2 , x](x − xi )(x − xi+1 )(x − xi+2 )dx
xi

4h5 4h5 f (4) (γi ) h5

=− f [xi , xi+1 , xi+2 , ζi , ζi ] = − · = − f (4) (γi ),
15 15 4! 90
for some γi ∈ [xi , xi+2 ].

20 / 25
 Simpson’s Rule
Truncation error cont.: Consequently,
Z xi+2
f [xi , xi+1 , xi+2 , x](x − xi )(x − xi+1 )(x − xi+2 )dx
xi

4h5 4h5 f (4) (γi ) h5

=− f [xi , xi+1 , xi+2 , ζi , ζi ] = − · = − f (4) (γi ),
15 15 4! 90
for some γi ∈ [xi , xi+2 ].

Total truncation error is

h5 (4)
| f (γ0 ) + f (4) (γ2 ) + · · · + f (4) (γn−2 ) |
90 | {z }
n numbers
2

h5 n f (4) (γ0 ) + f (4) (γ2 ) + · · · + f (4) (γn−2 ) (b − a)h4 (4)

= · = |f (γ)|
90 2 n/2 180
for some γ ∈ [a, b].

20 / 25
 Example 1
Rπ
Example: Compute 02 sin x dx by the Simpson’s rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 3 × 10−4 .

21 / 25
 Example 1
Rπ
Example: Compute 02 sin x dx by the Simpson’s rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 3 × 10−4 .

iπ π
Solution: Note that f (x) = sin x, xi = 8 for i = 0, . . . , 4, and h = 8.

21 / 25
 Example 1
Rπ
Example: Compute 02 sin x dx by the Simpson’s rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 3 × 10−4 .

iπ π
Solution: Note that f (x) = sin x, xi = 8 for i = 0, . . . , 4, and h = 8.
Z π   
2 1 π π π π 3π
sin x dx ≈ · sin 0 + sin + 2 sin + 4 sin + sin
0 3 8 2 4 8 8
= 1.0001.

21 / 25
 Example 1
Rπ
Example: Compute 02 sin x dx by the Simpson’s rule using n = 4
and round your answer to the nearest 4 decimal places. Show that
the truncation error is at most 3 × 10−4 .

iπ π
Solution: Note that f (x) = sin x, xi = 8 for i = 0, . . . , 4, and h = 8.
Z π   
2 1 π π π π 3π
sin x dx ≈ · sin 0 + sin + 2 sin + 4 sin + sin
0 3 8 2 4 8 8
= 1.0001.

(b − a)h4 (4)
Error = |f (γ)| for some γ ∈ [0, π2 ],
180
π π 4
( )
≤ 2 8 ≈ 2.08 × 10−4 < 3 × 10−4 .
180

21 / 25
 Example 2
R1 2
Example: Compute 0 e−x dx by Simpson’s rule so that the
truncation error is at most 10−4 . Round your answer to the nearest 4
decimal places.

22 / 25
 Example 2
R1 2
Example: Compute 0 e−x dx by Simpson’s rule so that the
truncation error is at most 10−4 . Round your answer to the nearest 4
decimal places.
2
Solution: We compute up to the 4th derivative of f (x) = e−x .
2
f ′ (x) = −2xe−x ,
2
f ′′ (x) = (4x 2 − 2)e−x ,
2
f ′′′ (x) = (12x − 8x 3 )e−x ,
2
f (4) (x) = (12 − 48x 2 + 16x 4 )e−x .

22 / 25
 Example 2
R1 2
Example: Compute 0 e−x dx by Simpson’s rule so that the
truncation error is at most 10−4 . Round your answer to the nearest 4
decimal places.
2
Solution: We compute up to the 4th derivative of f (x) = e−x .
2
f ′ (x) = −2xe−x ,
2
f ′′ (x) = (4x 2 − 2)e−x ,
2
f ′′′ (x) = (12x − 8x 3 )e−x ,
2
f (4) (x) = (12 − 48x 2 + 16x 4 )e−x .

Thus,
max |f (4) (x)| ≤ 76.
0≤x≤1

22 / 25
 Example 2
R1 2
Example: Compute 0 e−x dx by Simpson’s rule so that the
truncation error is at most 10−4 . Round your answer to the nearest 4
decimal places.
2
Solution: We compute up to the 4th derivative of f (x) = e−x .
2
f ′ (x) = −2xe−x ,
2
f ′′ (x) = (4x 2 − 2)e−x ,
2
f ′′′ (x) = (12x − 8x 3 )e−x ,
2
f (4) (x) = (12 − 48x 2 + 16x 4 )e−x .

Thus,
max |f (4) (x)| ≤ 76.
0≤x≤1

One can also use the tighter bound max0≤x≤1 |f (4) (x)| ≤ 20 (Why?). The resulting h and integral are left as an exercise.

22 / 25
 Example 2 cont.

Solution cont.: To force error ≤ 10−4 , it suffices to force

b−a 4
h max |f (4) (x)| ≤ 10−4 ,
180 0≤x≤1

which would be satisfied if

1 4
h · 76 ≤ 10−4 , which is guaranteed if h ≤ 0.1240.
180

23 / 25
 Example 2 cont.

Solution cont.: To force error ≤ 10−4 , it suffices to force

b−a 4
h max |f (4) (x)| ≤ 10−4 ,
180 0≤x≤1

which would be satisfied if

1 4
h · 76 ≤ 10−4 , which is guaranteed if h ≤ 0.1240.
180
Choose n = 10 so that h = 0.1, xi = 0.1i for i = 0, . . . , 10. Then
Z 1 " 4 5
#
−x 2 0.1 −02 −12
X
−(0.2i)2
X
−(0.2i−0.1)2
e dx ≈ e + e +2 e +4 e = 0.7468.
0 3
i=1 i=1

23 / 25
 Example 3
R1  
x2
Example: Compute 0
sin 2 + 1 dx by Simpson’s rule so that the
−4
truncation error is at most 10 . Round your answer to the nearest 4
decimal places.

24 / 25
 Example 3
R1  
x2
Example: Compute 0
sin 2 + 1 dx by Simpson’s rule so that the
−4
truncation error is at most 10 . Round your answer to the nearest 4
decimal places.
 2 
Solution: We compute up to the 4th derivative of f (x) = sin x2 + 1 .
 2   2   2 
x x x
f ′ (x) = x cos + 1 , f ′′ (x) = cos + 1 − x 2 sin +1 ,
2 2 2
 2   2 
x x
f ′′′ (x) = −x 3 cos + 1 − 3x sin +1 ,
2 2
 2   2 
(4) 2 x 4 x
f (x) = −6x cos + 1 + (x − 3) sin +1 .
2 2

24 / 25
 Example 3
R1  
x2
Example: Compute 0
sin 2 + 1 dx by Simpson’s rule so that the
−4
truncation error is at most 10 . Round your answer to the nearest 4
decimal places.
 2 
Solution: We compute up to the 4th derivative of f (x) = sin x2 + 1 .
 2   2   2 
x x x
f ′ (x) = x cos + 1 , f ′′ (x) = cos + 1 − x 2 sin +1 ,
2 2 2
 2   2 
x x
f ′′′ (x) = −x 3 cos + 1 − 3x sin +1 ,
2 2
 2   2 
(4) 2 x 4 x
f (x) = −6x cos + 1 + (x − 3) sin +1 .
2 2
Thus,
max |f (4) (x)| ≤ 10.
0≤x≤1

24 / 25
 Example 3 cont.
Solution cont.: To force error ≤ 10−4 , it suffices to force
b−a 4
h max |f (4) (x)| ≤ 10−4 ,
180 0≤x≤1

which would be satisfied if

1 4
h · 10 ≤ 10−4 , which is guaranteed if h ≤ 0.2059.
180

25 / 25
 Example 3 cont.
Solution cont.: To force error ≤ 10−4 , it suffices to force
b−a 4
h max |f (4) (x)| ≤ 10−4 ,
180 0≤x≤1

which would be satisfied if

1 4
h · 10 ≤ 10−4 , which is guaranteed if h ≤ 0.2059.
180
1 i
Choose 2n = 6so that h = 6 , xi = 6 for i = 0, . . . , 6. Then
R1
0
sin x2 + 1 dx ≈
  2   2 
1 0 1
sin + 1 + sin +1
18 2 2
2
! 3
!#
X ( 3i )2 X ( 2i−1
6 )
2
+2 sin +1 +4 sin +1 = 0.9091.
2 2
i=1 i=1

25 / 25