APPROXIMATE INTEGRATION

ELECTRONIC VERSION OF LECTURE

Dr. Lê Xuân Đại

HoChiMinh City University of Technology
Faculty of Applied Science, Department of Applied Mathematics
Email: ytkadai@hcmut.edu.vn

HCMC — 2017.

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 1 / 34

OUTLINE

1 PROBLEMS

2 TRAPEZOIDAL RULE

3 SIMPSON’S RULE

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 2 / 34

 Problems

PROBLEMS

There are two situations in which it is impossible to

find the exact value of a definite integral.
1
It is difficult, or even impossible, to find an
Z Z 1 1p
2
antiderivative. For example ex dx, 1 + x3 dx.
0 −1
2
The function is determined from a scientific
experiment through instrument readings or
collected data. There may be no formula for the
function.

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 3 / 34

 Trapezoidal Rule Riemann sum

RIEMANN SUM

Suppose the interval [a, b] is partitioned into n equal

b−a
parts, each of width ∆x = h = by the subdivision
n
points a = x0 , x1 = x0 + h, . . . , xk = x0 + kh, . . . , xn = b.
Then we have
Z b n
X
f (x)dx ≈ f (xi∗ )∆x, (1)
a i=1

where xi∗ is any point in the ith subinterval [xi−1 , xi ].

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 4 / 34

 Trapezoidal Rule Left endpoint approximation

If xi∗ is chosen to be the left endpoint of the interval,

then xi∗ = xi−1 and we have
Z b n
X
f (x)dx ≈ Ln = f (xi−1 )∆x (2)
a i=1

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 5 / 34

 Trapezoidal Rule Right endpoint approximation

If xi∗ is chosen to be the right endpoint of the

interval, then xi∗ = xi and we have
Z b n
X
f (x)dx ≈ Rn = f (xi )∆x (3)
a i=1

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 6 / 34

 Trapezoidal Rule Midpoint approximation

If xi∗ is chosen to be the midpoint of the interval,

xi−1 + xi
then xi∗ = and we have
2
Z b n ³x + x ´
i−1 i
∆x
X
f (x)dx ≈ Mn = f (4)
a i=1 2

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 7 / 34

 Trapezoidal Rule Trapezoidal Rule

TRAPEZOIDAL RULE

Let yk = f (xk ), k = 0, 1, . . . , n
hh i
The kth trapezoid has area f (xk−1 ) + f (xk ) .
2

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 8 / 34

 Trapezoidal Rule Trapezoidal Rule

So the sum of these areas of n trapezoids is

Zb Zx1 Zx2 Zxn
f (x)dx = f (x)dx + f (x)dx + . . . + f (x)dx
a x0 x1 xn−1

y0 + y1 y1 + y2 yn−1 + yn
≈ Tn = h · +h· +...+h·
2 2 2
h h i
= · y0 + 2y1 + 2y2 + . . . + 2yn−1 + yn
2

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 9 / 34

 Trapezoidal Rule Error Bounds

ERROR BOUNDS

DEFINITION 2.1
The error in using an approximation is defined to be
the amount that needs to be added to the
approximation to make it exact.
Z b
f (x)dx = approximation + error (5)
a

In general, we have
Z b Z b
ET = f (x)dx − Tn , EM = f (x)dx − Mn
a a

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 10 / 34

 Trapezoidal Rule Error Bounds

THEOREM 2.1
Suppose |f 00 (x)| É K , for a É x É b. If ET and EM are the
errors in the Trapezoidal and Midpoint Rules, then

K (b − a)3 K (b − a)3
|ET | É , |EM | É (6)
12n2 24n2

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 11 / 34

 Trapezoidal Rule Error Bounds

EXAMPLE 2.1
Use the Trapezoidal Rule and Left endpoint, Right
endpoint, Midpoint Rules with n = 5 to approximate
2
dx
Z
the integral ·
1 x

SOLUTION.
2−1 1 k 1 5
h= = , x0 = 1, xk = 1 + , yk = f (xk ) = =
5 5 5 1 + k5 5 + k
h P
µ ¶
4 1 P4 5 5
I≈ (yk + yk+1 ) = + ≈ 0.695635
2 k=0 20 k=0 5 + k 5 + (k + 1)

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 12 / 34

 Trapezoidal Rule Error Bounds

h³ ´
I ≈ y0 + 2y1 + 2y2 + 2y3 + 2y4 + y5
2

CASIO.
0.2
A = A+ × B × (1 ÷ X ) : X = X + 0.2
2
CALC A=0, X=1, B=1=.
A=, X=, B=2=.
...,...,...
A=, X=2, B=1=.
Answer: I ≈ 0.695635

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 13 / 34

 Trapezoidal Rule Error Bounds

MIDPOINT

h ³x + x ´ ³x + x ´ ³x + x ´
0 1 1 2 2 3
I ≈h f +f +f +
2 2 2
³x + x ´ ³x + x ´i
3 4 4 5
+f +f
2 2
CASIO.
A = A + 0.2 × (1 ÷ X ) : X = X + 0.2
CALC A=0, X=1.1.
A=, X=.
...,...,...
A=, X=1.9.
Answer: I ≈ 0.691908
Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 14 / 34
 Trapezoidal Rule Error Bounds

LEFT ENDPOINT

h i
I ≈ h f (x0 ) + f (x1 ) + f (x2 ) + f (x3 ) + f (x4 )
CASIO.
A = A + 0.2 × (1 ÷ X ) : X = X + 0.2
CALC A=0, X=1.
A=, X=.
...,...,...
A=, X=1.8.
Answer: I ≈ 0.745635

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 15 / 34

 Trapezoidal Rule Error Bounds

RIGHT ENDPOINT

h i
I ≈ h f (x1 ) + f (x2 ) + f (x3 ) + f (x4 ) + f (x5 )
CASIO.
A = A + 0.2 × (1 ÷ X ) : X = X + 0.2
CALC A=0, X=1.2.
A=, X=.
...,...,...
A=, X=2.
Answer: I ≈ 0.645635

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 16 / 34

 Trapezoidal Rule Error Bounds

2
dx
Z
Approximations to
1 x
n Ln Rn Tn Mn
5 0.745635 0.645635 0.695635 0.691908
10 0.718771 0.668771 0.693771 0.692835
20 0.705803 0.680803 0.693303 0.693069
2
dx
Z ¯2
The exact value = ln x¯ = ln 2.
¯
1 x 1
Corresponding errors
n EL ER ET EM
5 -0.052488 0.047512 -0.002488 0.001239
10 -0.025624 0.024376 -0.000624 0.000312
20 -0.012656 0.012344 -0.000156 0.000078
Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 17 / 34
 Trapezoidal Rule Error Bounds

EXAMPLE 2.2
UseZthe trapezoial rule with n = 10 to estimate
1
dx
I= ·
0 1+x

SOLUTION.
b−a 1−0 1 k
h= = = , x0 = 0, xk = ,
n 10 10 10
1 10
yk = f (xk ) = =
k 10 + k
1+
10
h P
µ ¶
9 1 P9 10 10
I≈ (yk + yk+1 ) = + ≈ 0.6938
2 k=0 20 k=0 10 + k 10 + (k + 1)

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 18 / 34

 Trapezoidal Rule Error Bounds

h³ ´
I ≈ y0 +2y1 +2y2 +2y3 +2y4 +2y5 +2y6 +2y7 +2y8 +2y9 +y10
2
CASIO.
0.1
A = A+ × B × (1 ÷ (1 + X )) : X = X + 0.1
2
CALC A=0, X=0, B=1=.
A=, X=, B=2=.
...,...,...
A=, X=1, B=1=.
Answer: I ≈ 0.6938

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 19 / 34

 Trapezoidal Rule Error Bounds

EXAMPLE 2.3
How large should we take n in order to guarantee that
the ZTrapezoidal and Midpoint Rule Approximations
2
dx
for are accurate to within 0.0001?
1 x
2
SOLUTION. f 00 (x) = 3 . So
x
2(2 − 1)3
¯ ¯
00
¯2¯
|f (x)| = ¯ 3 ¯ É 2, ∀x ∈ [1, 2] ⇒ |ET | É
¯ ¯ < 0.0001
x 12n2
1
⇒n> p ≈ 40.8 ⇒ n = 41.
0.0006
2(2 − 1)3 1
|EM | É < 0.0001 ⇒ n > p ≈ 28.86 ⇒ n = 29.
24n2 0.0012
Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 20 / 34
 Trapezoidal Rule Error Bounds

EXAMPLE 2.4
If the function f (x) is defined by table
x 1.0 1.2 1.4 1.6 1.8 2.0 2.2
then using
f (x) 4 3.3 2.4 4.3 10.2 6.2 7.4
the Trapezoidal Rule to approximate the integral
2.2
(xf 2 (x) + 4.4x3 )dx
R
I=
1.0

b − a 2.2 − 1.0
h= = = 0.2 ⇒ n = 6, x0 = 1.0, xk = 1.0 + 0.2k,
n n
yk = xk f 2 (xk ) + 4.4xk3 .

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 21 / 34

 Trapezoidal Rule Error Bounds

Therefore
h P
5 h
I≈ (yk + yk+1 ) = (y0 + 2y1 + 2y2 + 2y3 + 2y4 + 2y5 + y6 ).
2 k=0 2
CASIO h = 0.2, we have
h
A = A+ × B × (XY 2 + 4.4X 3 ) : X = X + h
2
CALC A=0, X=1.0, Y=4, B=1=.
CALC A=, X=, Y=3.3, B=2=.
...,...,...
CALC A=, X=2.2, Y=7.4, B=1=.
Answer. I ≈ 101.4579

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 22 / 34

 Simpson’s Rule

SIMPSON’S RULE
Another rule for approximate integration results
from using parabolas instead of straight line
segments to approximate a curve.
Suppose the interval [a, b] is partitioned into 2n equal
b−a
parts, each of width ∆x = h = by the subdivision
2n
points a = x0 , x1 = x0 + h, . . . , xk = x0 + kh, . . . , x2n = b. Let
yk = f (xk ), k = 0, 1, . . . , 2n

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 23 / 34

 Simpson’s Rule

We know that the equation of the parabola through

P0 , P1 and P2 is of the form y = A(x − x1 )2 + B(x − x1 ) + C
and so the area under the parabola from x0 = x1 − h to
x2 = x1 + h is
Z x2 h i Z h
2
A(x − x1 ) + B(x − x1 ) + C dx = (At 2 + Bt + C)dt
x0 −h

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 24 / 34

 Simpson’s Rule

Z h Z h
2
(At + Bt + C)dt = 2 (At 2 + C)dt =
−h 0
· 3 ¸h
x h
= 2 A + Cx = (2Ah2 + 6C)
3 0 3
But, since the parabola passes through P0 (x0 , y0 ),
P1 (x1 , y1 ), and P2 (x2 , y2 ), we have
y0 = A(−h)2 + B(−h) + C = Ah2 − Bh + C
y1 = C
y2 = Ah2 + Bh + C
and therefore y0 + 4y1 + y2 = 2Ah2 + 6C. Thus the area
h
under the parabola is (y0 + 4y1 + y2 ).
3
Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 25 / 34
 Simpson’s Rule

Similarly, the region under the curve y = f (x) on each

hh i
interval [xk , xk+2 ] has area f (xk ) + 4f (xk+1 ) + f (xk+2 ) .
3
So the sum of these areas of n regions is
Z b Z x2 Z x4 Z x2n
f (x)dx = f (x)dx + f (x)dx + . . . + f (x)dx
a x0 x2 x2n−2

h h h
≈ (y0 +4y1 +y2 )+ (y2 +4y3 +y4 )+..+ (y2n−2 +4y2n−1 +y2n ) =
3 3 3
h ³ ´
= y0 + 4y1 + 2y2 + 4y3 + 2y4 + . . . + 2y2n−2 + 4y2n−1 + y2n
3

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 26 / 34

 Simpson’s Rule

ERROR BOUNDS

THEOREM 3.1
Suppose |f (4) (x)| É K , for a É x É b. If ES is the error
involved in using Simpson’s Rule, then

K (b − a)5
|ES | É (7)
2880n4

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 27 / 34

 Simpson’s Rule

EXAMPLE 3.1
UseZSimpson’s rule with 2n = 20 to approximate
1
dx
I= and estimate the error involved in this
0 1+x
approximation.

SOLUTION.
b−a 1−0 1 k
h= = = , x0 = 0, xk = ,
2n 20 20 20
1 20
yk = f (xk ) = k
= ·
1 + 20 20 + k

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 28 / 34

 Simpson’s Rule

Therefore
h n−1
X
I≈ (y2k + 4y2k+1 + y2k+2 ) =
3 k=0
9 µ ¶
1 X 20 20 20
= +4 + ≈ 0.6931
60 k=0 20 + 2k 2k + 21 2k + 22
h³
I ≈ y0 +4y1 +2y2 +4y3 +2y4 +4y5 +2y6 +4y7 +2y8 +4y9 +2y10 +
3
´
+4y11 +2y12 +4y13 +2y14 +4y15 +2y16 +4y17 +2y18 +4y19 +y20

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 29 / 34

 Simpson’s Rule

CASIO.
B 1 1
A = A+ × :X =X+
6 × 10 X + 1 2 × 10
CALC A=0, B=1, X=0;
A=, B=4;X=;
A=, B=2;X=;
A=, B=4;X=;
A=, B=2;X=;
......
A=, B=1;X=1;
Answer. I ≈ 0.6931

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 30 / 34

 Simpson’s Rule

1 4!
The fourth derivative of f (x) = is f (4) (x) =
1+x (1 + x)4
⇒ |f (4) (x)| É 4!, ∀x ∈ [0, 1].

So we see that the error is at most

4!(1 − 0)5
4
≈ 8.33−7
2880 × 10

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 31 / 34

 Simpson’s Rule

EXAMPLE 3.2
Suppose the function f (x) is defined by table
x | 1.0 1.2 1.4 1.6 1.8 2.0 2.2
. Using
f (x) | 2 3.3 2.4 4.3 5.1 6.2 7.4
Simpson’s rule to approximate the integral
2.2
[xf 2 (x) + 2.2x3 ] dx.
R
I=
1.0

b − a 2.2 − 1.0
h= = = 0.2 ⇒ 2n = 6,
2n 2n
x0 = 1.0, xk = 1.0 + 0.2k.

yk = xk f 2 (xk ) + 2.2xk3 .

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 32 / 34

 Simpson’s Rule

0.2
I≈ (y0 + 4y1 + 2y2 + 4y3 + 2y4 + 4y5 + y6 )
3
CASIO.
0.2
A = A+B∗ ∗ (XY 2 + 2.2X 3 ) : X = X + 0.2
3
CALC A=0, B, X, Y is given by table
X | 1.0 1.2 1.4 1.6 1.8 2.0 2.2
Y | 2 3.3 2.4 4.3 5.1 6.2 7.4
B | 1 4 2 4 2 4 1

Therefore I = 59.82501333 ≈ 59.8250

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 33 / 34
 Simpson’s Rule

THANK YOU FOR YOUR ATTENTION

Dr. Lê Xuân Đại (HCMUT) APPROXIMATE INTEGRATION HCMC — 2017. 34 / 34