Numerical Integration & Differentiation Newton-Cotes Integration

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 29

Numerical Integration & Differentiation

Newton-Cotes Integration
Learning outcome
 Recognizing that Newton-Cotes integration formulas are based on the strategy of
replacing a complicated function or tabulated data with a polynomial that is easy to
integrate.

 Knowing how to implement the following single application Newton-Cotes


formulas:
 Trapezoidal rule
 Simpson’s 1/3 rule
 Simpson’s 3/8 rule

 Knowing how to implement the following composite Newton-Cotes formulas:


 Trapezoidal rule
 Simpson’s 3/8 rule
 Recognizing that even-segment-odd-point formulas like Simpson’s 1/3 rule achieve
higher than expected accuracy.
 Knowing how to use the trapezoidal rule to integrate unequally spaced data.
Introduction
 Differentiation?
 Process of finding derivative (dy/dx) - rate of change of a dependent variable
with respect to an independent variable.

Δy f(xi  Δx)  f(xi )



Δx Δx
dy f(xi  Δx)  f(xi )
 Δx lim 0
dx Δx
 Integration?
 Process of finding integral, I - the area under a function plotted on a graph.

f(x)

Area,A  I  f x  dx
b
a
A
Introduction
 Why? … engineering deals with systems/processes that are changing.
 derivative and integral calculus are used to describe physical worlds such as...
 Derivative - finding the velocity of a body from acceleration functions, and
displacement of a body from velocity data.
 Integral - finding area under the curves, centroid, moment of inertia, work
by variable force (Hooke’s Law), electric charge, average value, force by
liquid pressure.

 What can be differentiated and integrated?


1. Simple continuous function such as polynomial, exponential and trigonometry
– Analytical approach
2. Tabulated values (discrete data) – Numerical approach
3. Continuous complicated functions (difficult or impossible) – Numerical
approach.
Non-computer method for differentiation and integration
Equal-area graphical differentiation
 Given - (x,y) are tabulated data
a) Find divided difference (y/x) for each interval.
b) Plot y/x as stepped curve versus x
c) Draw smooth curve to approximate area under the curve.

Read from the


curve
Non-computer method for differentiation and integration
Using grid to approximate an integral
 Given - (x, y) are tabulated data.
a) Plot the curve
b) Count the number of boxes that approximate the area.

A = (heights x widths)

“better approximation is achieved using finer grid or thinner strips”


(Form the basis for computer methods)
Numerical integration
Numerical Integration (i.e. QUADRATURE)

Integral,I  A  f x  dx
b
a
f(x)

f(x) = integrand
x = variable
a, b = integration limits A
Methods of numerical integration
 Newton-Cotes integration formula
1. Trapezoidal rule
2. Simpson’s rule
 Simpson’s 1/3rule
 Simpson’s 3/8 rule

 Romberg integration
 Gauss quadrature
Newton-Cotes Integration Formula
 Newton-Cotes Formula;
 “Replace complicated function or tabulated data (i.e. integrand) with an nth
order polynomial across the integration interval”
b b

 
I  f(x)dx  f n (x)dx
a a

f n (x)  a 0  a1x    a n 1x n 1  a n x n Integrating function –


i.e. interpolating
polynomial
 fn(x) is an nth order interpolating polynomial.

 The integrating function can be


polynomials for any order, e.g.
a) straight lines – 1st order
b) parabolas – 2nd order
c) Cubic – 3rd order

Newton-Cotes closed formula


I = (b-a) x (Average Height)
Newton-Cotes Integration Formula
Methods of application

 Single application
 less accurate if a & b are widely spaced

 f x  dx
b
I
a
 Multiple-application
 repeatedly apply single integral for each
sub-interval of data points
 higher accuracy

b

I  f(x) dx  I1  I 2  I3  ....  I n
a

 Note: Higher order polynomial can be


used for higher accuracy
1.a Trapezoidal rule
 The trapezoidal rule uses a first order
polynomial as the integrating function.
b b

 
I  f(x)dx  f1 (x)dx
a a
f1 (x)  a 0  a 1x

b
 f1 (x)dx  Area of trapezoid
a

  Parallelsides 
  (Height) Newton-Cotes closed formula
 2  I = (b-a) x (Average Height)

 f(a)  f(b)  Trapezoidal for


 I  (b  a)  single integral
 2 
1.a Trapezoidal rule - Error
 the truncation error of a single application
of the trapezoidal rule is:

E t   f ξ b  a 
1 3

12
Where,  is somewhere between a
and b.

 error is dependent upon the second-


derivative of the actual function
 It is 1st order accurate  give exact result
if the actual function is linear.
 error is dependent on the third power of
the step size
 Error can thus be reduced by breaking the
curve into parts
  Use multiple-application/composite
1.a Trapezoidal rule - Example
Integrate f(x)=0.2 + 25x - 200x² + 675x3 - 900x4 + 400x5 from a=0 to b=0.8.
True value 1.640533.

f(0) = 0.2
f(0.8) = 0.232

 f(a)  f(b)   0.2  0.232 


I  (b  a)    ( 0.8  0)    0.1728
 2   2 

ε t  89.5%
1.b Multiple application of Trapezoidal rule
 A way to improve the accuracy of the trapezoidal
rule
 Sub-divide the interval [a, b] into n segment (sub-
interval)

ba
h a  x0 b  xn
n

 Apply single integral method to every segment,


total integral is
x1 x2 xn

I   f(x)dx   f(x)dx     f(x)dx


x0 x1 x n 1

 Substitute trapezoidal rule to each integral gives,


f(x 0 )  f(x1 ) f(x )  f(x 2 ) f(x )  f(x n )
Ih h 1    h n 1
2 2 2
1.b Multiple application of Trapezoidal rule
 Multiple-application Trapezoidal rule

h n 1

2 i 1

I  f(xo )  2 f(xi )  f(xn )

 Error = the sum of the individual error for each segment

(b  a) 3
Average value of second derivative
Ea   2
f  - Error is inversely related to n2
12n
 Problem
 For high accuracy, need more segments  more computations more
round-off errors!
 Solution
 Use higher-order polynomial as the integrating function!
1.b Multiple application of Trapezoidal rule - Example
Integrate f(x)=0.2 + 25x - 200x² + 675x3 - 900x4 + 400x5 from a=0 to b=0.8 with n=2.
True value 1.640533.

f(0) = 0.2
f(0.4) = 2.456
f(0.8) = 0.232

 0.2  2(2.456)  0.232 


I  (0.8  0)    1.0688
 4 

ε t  34.9%
2. Simpson’s Rules
 More accurate estimate of an integral is obtained if a high-order polynomial is used
to connect the points. The formulas that result from taking the integrals under such
polynomials are called Simpson’s rules.

a) Simpson’s 1/3 rule – use 2nd-order interpolating polynomial


b) Simpson’s 3/8 rule – use 3rd-order interpolating polynomial
2.a Simpson’s 1/3 Rule
 General form; b b

 
I  f(x)dx  f 2 (x)dx
a a

f 2 (x)  a 0  a 1x  a 2 x 2
 Using the Lagrange form for a quadratic fit of three points:

f n x  
x  x1  x  x 2  f x   x  x 0  x  x 2  f x   x  x 0  x  x1  f x 
x 0  x1  x 0  x 2  0 x1  x 0  x1  x 2  1 x 2  x 0  x 2  x1  2
 Integration over the three points simplifies to:
ba Sub-interval is
x2
f 2 x  dx h

equally spaced
I 2
x0

I
h
f x 0   4f x1   f x 2  simpson’s 1/3 rule
3 for single integral
Where, x0 = a , x1=(a+b)/2 & x2 = b
Newton-Cotes closed formula

 Called 1/3 because h is divided by 3 I = (b-a) x (Average Height)


2.a Simpson’s 1/3 Rule - Error
 Truncation error of a single application of Simpson’s 1/3 rule is:

Et 
b  a  4 
5
f ξ 
2880
where  is somewhere between a and b.

 error is dependent upon the fourth-derivative of the actual function

 Simpson’s 1/3 is third-order accurate though it is derived from 2nd order polynomial
more accurate than expected! Give exact result for cubic polynomial.

 error is dependent on the fifth power of the step size (rather than the third for the
trapezoidal rule).
2.b Multiple-Application Simpson’s 1/3 Rule
 A way to improve the accuracy of the Simpson’s
1/3 rule

 Sub-divide the interval [a,b] into n segment (sub-


interval)
ba
h ax bx
n 0 n
 Apply single integral method over every two
segment  n must be even!, total integral is
x2 x4 xn

I  f(x)dx   f(x)dx    f(x)dx


x0 x2 x n 2

 Substitute Simpson’s 1/3 rule to each integral


gives

I
h
f x 0   4f x1   f x 2   h f x 2   4f x 3   f x 4     h f x n 2   4f x n 1   f x n 
3 3 3
2.b Multiple-Application Simpson’s 1/3 Rule
 Multiple-application Simpson’s 1/3 rule;
 n 1 n 2

h 

I  f x 0   4 f x i   2
3

f x i   f x n 
 i 1 j 2 
 i, odd j, even 
 Error = the sum of the individual error for each segment

(b  a ) 5 Average value of 4th derivative


Ea   f (4)
180n 4
 Very accurate  superior than Trapezoidal

 Problem
 Limited to only equi-spaced data!
 Must have even number of segments (i.e. odd number of data points)
 Solution
 For odd number of segment s  use Simpson’s 3/8 rule!
2.b Multiple-Application Simpson’s 1/3 Rule
 Multiple-application Simpson’s 1/3 rule;
 n 1 n 2

h 

I  f x 0   4 f x i   2
3

f x i   f x n 
 i 1 j 2 
 i, odd j, even 
 Error = the sum of the individual error for each segment

(b  a ) 5 Average value of 4th derivative


Ea   f (4)
180n 4
 Very accurate  superior than Trapezoidal

 Problem
 Limited to only equi-spaced data!
 Must have even number of segments (i.e. odd number of data points)
 Solution
 For odd number of segment s  use Simpson’s 3/8 rule!
2.c Simpson’s 3/8 Rule
 General form;
b b

 
I  f(x)dx  f 3 (x)dx
a a

f 3 (x)  a 0  a1x  a 2 x 2  a 3 x 3
 Using the Lagrange form for a 3rd-order polynomial to fit of four points:
 Integration over the three points simplifies to:
Sub-interval is
ba
 f x  dx h
x2 equally spaced
I 3
x0 3
I  f x 0   3f x1   3f x 2   f x 3 
3h SIMPSON’S 3/8 RULE
8 for single integral

Where, x0 = a , x1=(a+b)/3, x2=(2a+2b)/3 & x2 = b


2.c Simpson’s 3/8 Rule
 Truncation error of a single application of Simpson’s 3/8 rule is:

Et 
b  a  4 
5
f ξ 
6480
where  is somewhere between a and b.

 error is dependent upon the fourth-derivative of the actual function

 Simpson’s 3/8 is still third-order accurate though it is derived from 3rd order
polynomial  Give exact result for cubic polynomial.

 error is dependent on the fifth power of the step size


2.c Simpson’s 3/8 Rule
REMARKS

 Simpson’s 1/3 is usually the preferred


method because it attains third-order
accuracy with 3 points rather than the
4 points required for the 3/8 version.

 Simpson’s 3/8 rule is generally used in


tandem with Simpson’s 1/3 rule when
the number of segments is odd.
2.d Multiple-Apllication Simpson’s 3/8 Rule
 Sub-divide the interval [a, b] into n segment (sub-
interval)
ba
h a  x0 b  xn
n
 Apply single integral method over every THREE
segment  n must be divisible by three!, total
integral is
x3 x6 xn

 
I  f(x)dx  f(x)dx   
x0 x3
 f(x)dx
x n 3

 Substitute Simpson’s 3/8 rule to each integral gives,

3h  n 1 n 2 
I f x 0   3
8 
 (f x i   f x i 1 )  2
i  2,5,7..
 f x i   f x n 
j 4,7,10... 
2. Simpson’s Rule
FINAL REMARKS

 Simpson’s 1/3
 for EVEN number of segments
 3rd order accuracy (although based on 2nd order polynomial) thus it is a
preferred method!

 Simpson’s 3/8
 For ODD number of segments & segment must be divisible by 3.
 Can be used in combination with 1/3 rule to solve for ODD number of
segments

 Simpson’s rules are sufficient for most applications.

 Higher-order Newton-Cotes formulas may also be used - in general, the higher the
order of the polynomial used, the higher the derivative of the function in the error
estimate and the higher the power of the step size. more accurate but rarely
used
2. Simpson’s Rule
FINAL REMARKS

 Simpson’s 1/3
 for EVEN number of segments
 3rd order accuracy (although based on 2nd order polynomial) thus it is a
preferred method!

 Simpson’s 3/8
 For ODD number of segments & segment must be divisible by 3.
 Can be used in combination with 1/3 rule to solve for ODD number of
segments

 Simpson’s rules are sufficient for most applications.

 Higher-order Newton-Cotes formulas may also be used - in general, the higher the
order of the polynomial used, the higher the derivative of the function in the error
estimate and the higher the power of the step size. more accurate but rarely
used
2. Simpson’s Rule
 For unequal segments, one method is to
 apply trapezoidal rule to each segment and
 sum the results

f(x0 )  f(x1 ) f(x1 )  f(x2 ) f(xn 1 )  f(xn )


I  h1  h2  hn
2 2 2

 Where hi = the width of segment i

 Same approach as multiple-application trapezoidal rule


but h is the width of the unequal segments.
Newton-Cotes integration formula – Final remark
 Numerical integration for;
1. Tabulated values (discrete data).
2. Continuous complicated functions (difficult or impossible).

 Concept:
 “Replace tabulated data and complicated function (i.e. integrand) with an nth
order integrating polynomial across the integration interval”
b b
I   f(x)dx   f n (x)dx
a a

 Accuracy;
 Higher order polynomial  more accurate

 Method of application:
 Single application  not accurate
 Multiple application  more accurate (need more data values!)

You might also like