Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

CHAPTER 3
NUMERICAL INTEGRATION
AND DIFFERENTIATION
Dr. Le Thanh Long
[email protected]

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

OUTLINE
3.1. Numerical Integration
3.2. Numerical Differentiation
3.3. Lagrange Interpolating Polynomials

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

Figure 3.1: Examples of how integration is used to evaluate areas in engineering

applications.
(a) Evaluating the area of a field bounded by a meandering stream and two roads.
(b) Evaluating the cross-sectional area of a river.
(c) Determining the net force due to a nonuniform wind blowing against the side of
a skyscraper.
3

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

Newton-Leibnitz Integration Formulas
b b
 f ( x)dx  F ( x)  F (b)  F (a)
a a
where F’(x) = f(x), F is a primitive function of f.

But we have often to determine the integration of the

function y = f(x) defined by the data table. So, the concept
of primitive function makes sense no longer.

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

Newton-Cotes Integration Formulas
• This is the most common numerical integration schemes.
• Base on the strategy of replacing a complicated function
or tabulated data with an approximating function that is
easy to integrate:
b b
I   f ( x)dx   f n ( x)dx
a a
where:

f n ( x)  a0  a1 x  a2 x 2  ...  an x n

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

Example 3.1: Computing the integration by analytical method (Newton-
Leibnitz) and numerical integration method (Newton-Cotes)
1
2
 (96 x  42 x  13) dx
Solution: 1

Let I1 the integration calculated by analytical method (Newton-Leibnitz) và I2

the integration calculated by numerical method (Newton-Cotes)
- Analytical method:
1
2 3 2 1
I1   (96 x  42 x  13) dx  (32 x  21x  13 x )  38
1
1
- Numerical method:
Look up Table 1.1 in Appendix 1, the abscissa of the integral points xi and the
corresponding weights ωi are x1 = -1, ω1 = ; x2 = 0, ω2 = ; x3 = 1, ω3 =
1
1 4 1
I 2   (96 x2  42 x 13)dx  [96(1)2  42(1) 13]  [96(0)2  42(0) 13]  [96(1)2  42(1) 13]
1
3 3 3
41 52 125
    38
3 3 3 5

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

1. The Trapezoidal Rule
Using first order Taylor series to approximate ,
b b
I   f ( x)dx   f1 ( x)dx
a a

where:
f (b)  f (a )
f1 ( x)  f (a )  ( x  a)
ba
Then:
b f (b)  f (a )  f (b)  f (a)
I    f (a)  ( x  a)   (b  a)
a
 ba  2

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

1. The Trapezoidal Rule
The trapezoidal rule is
equivalent to approximating
the area of the trapezoidal
under the straight line
connecting ( ) and .
An estimated for the local
truncation error of a single
application of the
trapezoidal rule is:
Figure 3.2: Graphical depiction of 1
(b  a )3 f "( x)
the trapezoidal rule. 12

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

1. The Trapezoidal Rule
Example 3.2: Use the trapezoidal rule to numerically integrate the function
f(x) = 0.2 + 25x from a = 0 to b = 2
Solution:
2
2
2
The true solution is  f ( x)dx  (0.2  12.5 x )  50.4
0 0
f (a )  f (0)  0.2
f (b)  f (2)  50.2

f (b)  f (a ) 0.2  50.2

I  (b  a )  (2  0)  50.4
2 2
Because ( ) is a linear function, using the trapezoidal rule gets the
exact solution.
8

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

1. The Trapezoidal Rule
Example 3.3: Use the trapezoidal rule to numerically integrate the function
f(x) = 0.2 + 25x + 3x2 from a = 0 to b = 2
Solution: f (0)  0.2; f (2)  62.2
f (b)  f ( a )
I  (b  a )  62.4
2
The exact solution is
2
2 3 2
 f ( x)dx  (0.2 x  12.5 x  x )  58.4
0 0
The relative error is:
58.4  62.4
t  100%  6.85%
58.4
9

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

1. The Trapezoidal Rule
Multiple-application trapezoidal rule
Using a smaller integration interval can reduce the approximation
error. Divide [a, b] into n segments of equal width h = .
n 1
f ( x0 )  2 i 1 f ( xi )  f ( xn )
I  (b  a )
2n
Error:
(b  a )3
Ea   2
f "( x)
12n

10

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

1. The Trapezoidal Rule
Example 3.4: Use the 2-segments trapezoidal rule to numerically
integrate and estimate the error of f(x) = 0.2 + 25x + 3x2 from a = 0 to b = 2
Solution: (b  a)
n  2; h  1
n
f (0)  0.2; f (1)  28.2; f (2)  62.2
f (0)  2 f (1)  f (2)
I  (b  a)  59.4
2n
Error: 2
 6dx
0
f "( x)  6, f "( x)  6
3
20
(2  0)
Ea   2
6  1
12x2
The exact solution is 59.4 – 1 = 58.4 The relative error is 1.71%
11

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

2. Simpson’s Rule
Aside from applying the trapezoidal rule with finer segmentation,
another way to improve the estimation accuracy is to use higher-
order polynomials.
Multiple-application Simpson’s rule:
b m m 1
h 
 f ( x)dx   f 0  f 2 m  4 f 2 k 1  2 f 2 k 
a
3 k 1 k 1 
(b  a )5 (4)
Error: Ea   4
f
180n

Where ̅( ) is the average fourth derivative for the interval.

13

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

2. Simpson’s Rule
Example 3.5: With n = 4, estimate the integral and error of
= 0.2 + 25 − 200 + 675 − 900 + 400
From a = 0 to b = 0.8. The exact integral is 1.640533
Solution: h = 0.2
f(0) 0.2 0.2
I  0.2  4(1.288  3.464)  2(2.456)  0.232  1.623467
3
f(0.2) 1.288
f(0.4) 2.456 = 1.640533 − 1.623467 = 0.017067
f(0.6) 3.464 = 1.04%
f(0.8) 0.232 0.85
The estimated error is Ea  4
(2400)  0.017067
180(4)

14

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.1. Numerical Integration

2. Simpson’s Rule
Example 3.6: Use Simpson’s rule to estimate the integral and error
of 0 1 dx with the number of segments n = 6
0.6

1 x
0, 6  0
Solution: h  0,1
6
0.1
x y I { f (0)  f (0.6)  4[ f (0.1)  f (0.3)  f (0.5)]
3
0 (0) = 1  2[ f (0.2)  f (0.4)]}  0.470
0.1 (0.1) = 0.909 0.6 24
0.2 (0.2) = 0.833 24 0 (1  x)5 dx
f ""( x)  5
; f ""( x)   8.47
(1  x) 0.6  0
0.3 (0.3) = 0.769
0.4 (0.4) = 0.714 0.65
Ea   4
8.47  0.000003
180  6
0.5 (0.5) = 0.667
0.6 (0.6) = 0.625 => True integral ≈ 0.469997
15

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

Problem 1. Evaluate the following integral:

 /2

 (6  3cos x)dx
0

a) Multiple-application trapezoidal rule, with n = 4.

b) Multiple-application Simpson’s rule, with n = 4.

c) Determine the percent relative error of the two methods above.

16

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

Problem 1.
Solution:
a) Multiple-application trapezoidal rule

−0
ℎ=2 = /8
4

= 0; = /8; = /4; = 3 /8; = /2

 f (0)  f ( / 2)  2[ f ( / 8)  f ( / 4)  f (3 / 8)]

I
8 2
  3 
 [9  6  2[18  3cos  3cos  3cos ]
16 8 4 8
 12.386
17

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

Problem 1.
Solution:
b) Multiple-application Simpson’s rule
−0
ℎ=2 = /8
4
= 0; = /8; = /4; = 3 /8; = /2

 f (0)  4[ f ( / 8)  f (3 / 8)]  2 f ( / 4)  f ( / 2)

I
8 3

 [9  4(12  3cos( / 8)  3cos(3 / 8))  2(6  3cos( / 4))  6]  12.4252
24

18

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

Problem 1.
Solution:
c) Determine the percent relative error
 /2  /2  /2

 (6  3cos x)dx  6  dx  3  cos xdx

0 0 0

 3  3  12.42478
Multiple-application trapezoidal rule
12.386  12.42478
T  100%  0.312%
12.42478
Multiple-application Simpson’s rule

12.4252  12.42478
T  100%  0.003%
12.42478
19

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.2. Numerical Differentiation

1. First Order Derivatives

Figure 3.3: Graphical depiction of forward, backward and centered

finite-divided-difference approximations of the first derivative. 20

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.2. Numerical Differentiation

1. First Order Derivatives
The first forward finite divided difference is
f ( xi 1 )  f ( xi )
f '( xi ) 
h
The first backward finite divided difference is
f ( xi )  f ( xi 1 )
f '( xi ) 
h
The first centered finite divided difference is
f ( xi 1 )  f ( xi 1 )
f '( xi ) 
2h
In case that we consider 4 points
 f  x0  2h   8 f  x0  h   8 f  x0  h   f  x0  2h 
f   x0  
12h 21

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.2. Numerical Differentiation

Example 3.7: Estimate the first order derivative of f(x) at x = 0.5 and
h = 0.5
f ( x)  0.1x 4  0.15 x 3  0.5 x 2  0.25 x  1.2
Solution:
 xi  0.5  f ( xi )  0.925
 
 xi 1  xi  h  0   f ( xi 1 )  1.2
x  x  h  1  f ( x )  0.2
 i 1 i  i 1

The first forward divided difference:

f ( xi 1 )  f ( xi )
f '( xi )   1.45
h
The first backward divided difference:
f ( xi )  f ( xi 1 )
f '( xi )   0.55
h 22

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.2. Numerical Differentiation

Example 3.7
Solution:
The first centered divided difference:
f ( xi 1 )  f ( xi 1 )
f '( xi )   1.0
2h

Exact differentiate: -0.9125

=> Using centered finite divided difference and small step size
achieves lower approximation error

23

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.2. Numerical Differentiation

2. Second Order Derivatives
The second forward finite divided difference is
f ( xi  2 )  2 f ( xi 1 )  f ( xi )
f "( xi ) 
h2
The second backward finite divided difference is
f ( xi )  2 f ( xi 1 )  f ( xi  2 )
f "( xi ) 
h2
The second centered finite divided difference is
f ( xi 1 )  2 f ( xi )  f ( xi 1 )
f "( xi ) 
h2
In case that we consider 5 points
 f  x0  2h   16 f  x0  h   30 f  x0   16 f  x0  h   f  x0  2h 
f   x0  
12h 2 24

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.2. Numerical Differentiation

2. Second Order Derivatives
Example 3.8: Evaluate the second derivatives of function f(x) at x = 0.5
and h = 0.25
f ( x)  0.1x 4  0.15 x3  0.5 x 2  0.25 x  1.2
Solution:
 xi  0.5  f ( xi )  0.925
 x  0.25  f ( x )  1.1035 Using the forward f.d.d: f "( xi )  2.326
 i 1  i 1

 xi  2  0   f ( xi  2 )  1.2 Using the backward f.d.d: f "( xi )  1.312

 x  0.75  f ( x )  0.6363
 i 1  i 1 Using the centered f.d.d: f "( xi )  1.7632
 xi  2  1  f ( xi  2 )  0.2
Exact differentiate: -1.75

25

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

Problem 2. Use the following data to find the velocity and

acceleration at t = 10 seconds. Use centered finite difference.

Time, t, s 0 2 4 6 8 10 12 14 16
Position, x, m 0 0.7 1.8 3.4 5.1 6.3 7.3 8.0 8.4

26

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

Solution:

• Velocity: f ( xi 1 )  f ( xi 1 )
f '( xi ) 
2h

f ( x6 )  f ( x4 )
f '( x5 ) 
2h
f (12)  f (8) 7.3  5.1
f '(10)    0.55(m / s )
2 2 2 2
• Acceleration:
f ( xi 1 )  2 f ( xi )  f ( xi 1 )
f "( xi ) 
h2
f ( x6 )  2 f ( x5 )  f ( x4 )
f "( x5 ) 
h2
f (12)  2 f (10)  f (8) 7.3  2  6.3  5.1 2
f "(10)  2
 2
 0.05( m / s )
2 2 27

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

Problem3. Determine y’(1) and y”(1) of = ,

with h = 0.1.

Results:
1 = −0.17824017
"(1) = 0.3573462

28

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.3 Lagrange Interpolating Polynomials

The Lagrange interpolating polynomial is simply a reformulation
of the Newton polynomial that avoids the computation of divided
differences. It can be represented concisely as
n
f n ( x)   Li ( x) f ( xi )
i 0
where
n x  xj
Li ( x)  
j 0 xi  x j
j i

29

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

3.3 Lagrange Interpolating Polynomials

Example 3.9: Use a Lagrange interpolating polynomial of the first
and second order to evaluate f(x) = ln(x) at x = 2 on the basis of the
data given as:
x0  1 f ( x0 )  0
x1  4 f ( x1 )  1.386294
x2  6 f ( x2 )  1.791760
Solution:
The first-order polynomial can be used to obtain the estimate at x = 2
24 2 1
f1 (2)  0 1.386294  0.4620981
1 4 4 1
Similarly, the second-order polynomial is developed as:
(2  4)(2  6) (2  1)(2  6) (2  1)(2  4)
f 2 (2)  0 1.386294  1.791760  0.5658444
(1  4)(1  6) (4  1)(4  6) (6  1)(6  4)
30

Department of Machine Design - Faculty of Mechanical Engineering

Ho Chi Minh City University of Technology (HCMUT), VNU-HCM

31

Department of Machine Design - Faculty of Mechanical Engineering