Kuliah5.1 - (Integrasi Numerik)

Numerical Integration
Roger Crawfis
Outline
• Definite Integrals
• Lower and Upper Sums
– Reimann Integration or Reimann Sums
• Uniformly-spaced samples
– Trapezoid Rules
– Simpson’s Rules
– Adaptive Simpson’s Scheme
• Non-uniformly spaced samples
– Gaussian Quadrature Formulas
February 4, 2020 OSU/CSE 541 2

Motivation
What does an integral represent?
d b
 f ( x)dxdy  volume
b
a
f ( x)dx  area c a
Basic definition of an integral: f(x)

n
f ( x )dx  lim  f ( xk )x
b
a n 
k 1
where x  b  a
n
sum of height  width x

Motivation
b
• Evaluate the integral, I   a
f ( x)dx without doing the
calculation analytically.
• Necessary when either:
– Integrand is too complicated to integrate analytically
2 2  cos(1  x ) 0.5 x
0
1  0.5 x
e dx
– Integrand is not precisely defined by an equation, i.e.,

we are given a set of data (xi, yi), i = 1, 2, 3, …,n
Reimann Integral Theorem
• Integration is a summing process. Thus virtually
all numerical approximations can be represented
by n
I   f ( x )dx   wi f ( xi )  Et
b
a
i 1
in which wi are the weights, xi are the sampling

points, and Et is the truncation error
• Valid for any function that is continuous on the
closed and bounded interval of integration.

Partitioning the Integral
• The most common numerical integration formula
is based on equally spaced data points.
xn
 x0
f ( x)dx
• Divide [x0 , xn] into n intervals (n1)

x1 x2 xn
xn
 x0
f ( x)dx   f ( x)   f ( x) 
x0 x1
 
xn1
f ( x)

Upper Sums
• Assume that f(x)>0 everywhere.

• If within each interval, we could determine
the maximum value of the function, then we
have: xn n 1
 f ( x)   M  x
x0 i 0
i i 1  xi 
• where
M i  sup f ( x) : xi  x  xi 1 Supremum - least
upper bound

Upper Sums
• Graphically:
x0 x1 x2 x3 x4

Lower Sums
• Likewise, still assuming that f(x)>0

everywhere.
• If within each interval, we could determine
the minimum value of the function, then we
have: f ( x)  m x  x
xn n 1
  
i 0
i i 1 i 
x0
• where
mi  inf  f ( x) : xi  x  xi 1
Infimum - greatest
lower bound

Lower Sums
• Graphically
x0 x1 x2 x3 x4

Finer Partitions
• We now have a bound on the integral of the
function for some partition (x0,..,xn):
n 1 xn n 1
m x
i 0
i i 1  xi    f ( x)   M i  xi 1  xi 
i 0
x0
• As n, one would assume that the sum of the

upper bounds and the sum of the lower bounds
approach each other.
• This is the case for most functions, and we call
these Riemann-integrable functions.

Bounding the Integral
• Graphically
x0 x1 x2 x3 x4

• Halving each interval (much like Lab1):
x0 x3 x5 x7 x9

• One more time:
x0 x5 x7 x9 x11

Monotonic Functions
• Note that if a function is monotonically

increasing (or decreasing), then the lower
sum corresponds to the left partition values,
and the upper sum corresponds to the right
partition values.
x0 x3 x5 x7 x9
Lab1 and Integration
• Thinking back to lab1, what were the limits

or the integration?
• Is the sin function monotonic on this
interval?
• Should the Reiman sum be an upper or
lower sum?

Polynomial Approximation
• Rather than search for the maximum or minimum,
we replace f(x) with a known and simple function.
• Within each interval we approximate f(x) by an mth
order polynomial.
pm ( x)  a0  a1 x  a2 x 2  ...  am x m

Newton-Cotes Formulas
• The m’s (order of the polynomials) may be the
same or different.
xn x0m1 x0m1m2 xn

x0
f ( x)dx  
x0
pm1 ( x)dx  
x0m1
pm2 ( x)dx  ...  
xnmn
pmn ( x)dx
• Different choices for m’s lead to different

formulas: m Polynomial Formula Error
1 linear Trapezoid O(h 2 )
2 quadratic Simpson' s 1/3 O(h 4 )
3 cubic Simpson' s 3/8 O(h 4 )
Trapezoid Rule
• Simplest way to approximate the area under a
curve – using first order polynomial (a straight
line)
• Using Newton’s form of the interpolating
polynomial: f b   f a 
p1  x   f a   x  a 
ba
• Now, solve for the integral:
I   f x dx   p1 x dx
b b
a a

Trapezoid Rule
 f b   f a  
 f a   x  a dx
b
I 
a ba 
Trapezoid Rule
f(b)
I
b  a
 f a   f b
f(a)
2
I  width  average height

a b
Trapezoid Rule
• Improvement?
x0 x1 x2 x3 x4

Trapezoid Rule Error
• The integration error is:
(b  a )
f  h 3   f  h 2
1
Et  
12 12
O(h 3 )
• Where h = b - a and  is an unknown point where
a <  < b (intermediate value theorem)
• You get exact integration if the function, f, is

linear (f = 0)

Example
 x2
Integrate from f ( x)  e a = 0 to b = 2
Use trapezoidal rule:
2
I   e dx  x2
0

b  a
 f a   f b  (2  0)
 f (2)  f (0)
2 2
 1  (e 4  e0 )  1.0183

Example
Estimate error: Et   1 f  h 3

12
Where h = b - a and a <  < b
Don’t know  - use average value

f ( x)  (2  4 x )e
2  x2 f (0)  2
h  20  2 f ( 2)  0.2564
23  f 0  f 2
Et  Ea    0.58
12 2
More intervals, better result [error  O(h2)]
7 7
6
n=2 6
n=3
5 5
4 4
3 3
2 2
1 1
0 0
3 5 7 9 11 13 15 3 5 7 9 11 13 15
7 7
6
n=4 6
n=8
5 5
4 4
3 3
2 2
1 1
0 0
3 5 7 9 11 13 15 3 5 7 9 11 13 15

Composite Trapezoid Rule
• If we do multiple intervals, we can avoid duplicate
function evaluations and operations:
• Use n+1 equally spaced points.
• Each interval has: h  b n a
• Break up the limits of integration and expand.
a h a 2 h
f x dx   f x dx  ...   f x dx
b
I 
a a h bh

• Substituting the trapezoid rule for each integral.
ah a2h
I   f  x  dx  ah f  x  dx  ...  bh f  x  dx
b
a  h  a  a  2h  a  h 
  f  a   f  a  h    f  a  h   f  a  2h  
2 2
b  b  h 
....   f  b  h   f  b 
2
• Results in the Composite Trapezoid Formula:

h n 1

I   f a   2 f a  ih   f b 
2 i 1 
• Think of this as the width times the average

height.
h n 1

I   f a   2 f a  ih   f b 
2 i 1 
n 1
f a   2 f a  ih   f b 
 b  a  i 1
2n
width
Average height

Error
• The error can be estimated as:


b  a h 2
f  
b  a
3
f 
2
Ea
12 12n 2 O(h )
• Where, f  is the average second derivative.

• If n is doubled, h  h/2 and Ea  Ea/4
• Note, that the error is dependent upon the
width of the area being integrated.
Example
• Integrate: f x   0.3  20 x  140 x 2  730 x 3  810 x 4  200 x5
• from 40
a=0.2 35
to 30
b=0.8 25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Example
• A single application of the Trapezoid rule.

f a   f b 
I  b  a 
2
34.22  3.81
 0.8  0.2
2
 11.26
• Error:
f  b  a 
1
Et  
3
12

Example
• We don’t know  so approximate with

average f
f x   20  280 x  2190 x 2  3240 x 3  1000 x 4
f x   280  4380 x  9720 x 2  4000 x 3

0.8
f  x  

0.2
f  dx
0.8  0.2
f (0.8)  f (0.2)
  131.6
0.8  0.2
Example
• The error can thus be estimated as:

 b  a h  b  a
2 3
Et  f   2
f 
12 12n
1
   131.6  0.8  0.2   2.37
3
12

True value of integral is 12.82. Trapezoid
rule is 11.26 - within approx error - Et is 12%
40
35
30
25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2

Using Three Intervals
• Use intervals (0.2,0.4),(0.4,0.6),(0.6,0.8):

– (n = 3, h = 0.2)
n 1
f a   2 f a  ih   f b 
I  b  a  i 1
2n
f 0.2   2 f 0.4   f 0.6   f 0.8
 0.8  0.2 
( 2)( 3)
3.31  213.93  30.16  34.22
 0.6
6
 12.57
True value of integral is 12.82
Et is now 2%
40
35
30
25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2

Using Six Intervals
• Use intervals (0.2,0.3),(0.3,0,4), etc.

– (n = 6, h = 0.1)
f 0.2   2 f 0.3  f 0.4   f 0.5  f 0.6  f 0.7   f 0.8

I  0.8  0.2 
(2)( 6)
3.31  27.34  13.93  22.18  30.16  35.22   34.22
 0.6
12
 12.76
True value of integral is 12.82

Et is now 0.5%
40
35
30
25
20
15
10
0
0 0.2 0.4 0.6 0.8 1 1.2

Multi-dimensional Integration
• Consider a two-dimensional case.

i 
1 1 1 n
 f  x, y  dxdy    Ai f  , y  dy
0 0 0 i 0 n 
i 
n 1
  Ai  f  , y  dy
i 0 0 n 
n
 i j n
  Ai  Aj f  , 
i 0 j 0 n n
n
 i j
n
  Ai Aj f  , 
i 0 j 0 n n
• For the Trapezoid Rule, this leads to

weights in the following pattern:
1 1 2 2 2 2 2 1
2 2 4 4 4 4 4 2 1 i  0, n j  0, n

2 2 4 4 4 4 4 2 1 2 i  1, , n  2 j  1, n  1
Aij  2 
2 2 4 4 4 4 4 2
4n  2 i  {1, n  1} j  1, , n  1
4 i  1, , n  2 j  1, , n  2
2 2 4 4 4 4 4 2
2 2 4 4 4 4 4 2 
1 1 2 2 2 2 2 1
1 2 2 2 2 2 1

• If we use the weights from the Trapezoid

rule, the error is still O(h2).
• However, there are now n2 function
evaluations.
– Equally-spaced samples on a square region.

• In general, given k dimensions, we have
N= nk function evaluations:
 2
   
2
O  h2   O  n2   O   nk  k   O  N k 

   
• If the dimension is high, this leads to a
significant amount of additional work in
going from hh/2.
– Remember this for Monte-Carlo Integration.

Reducing the Error
• To improve the estimate of the integral, we

can either:
– Add more intervals
– Use a higher order polynomial
– Use Richardson Extrapolation to examine the
limit as h0.
• Called Romberg Integration

Adding More Intervals
• If we have an estimate for one value of h,

do we need to recompute everything for a
value of h/2?
h n 1

I h   f  a   2 f  a  ih   f  b  
2 i 1 

Adding More Intervals
• This is called the Recursive Trapezoid

Rule in the book.
• We have n 2n and hh/2.
h 2 n 1
 h 
I h   f (a )  2  f  a  i   f (b) 
2 4 i 1  2 
h n 1 n 1
 h 
  f (a)  2 f  a  ih   2 f  a  ih    f (b) 
4 i 1 i 0  2 
I h h  n 1  h 
   2 f  a  ih   
2 4  i 0  2 
Higher-Order Polynomials
• Recall:
xn x0m1 x0m1m2 xn

x0
f ( x)dx  
x0
pm1 ( x)dx  
x0m1
pm2 ( x)dx  ...  
xnmn
pmn ( x)dx
m Polynomial Formula Error

1 linear Trapezoid O(h 2 )
2 quadratic Simpson' s 1/3 O(h 4 )
3 cubic Simpson' s 3/8 O(h 4 )

Simpson Rule
Simpson’s 1/3 Rule
• If we use a 2nd order polynomial (need 3

points or 2 intervals):
 x  x2 
– Lagrange form.  x1  0 
 2 
  x  x1  x  x2  x  x0 x  x2  f x 
f  x0  
x2
I  
x0  x  x  x  x       1
 0 1 0 2 x1 x0 x1 x 2

 x  x0  x  x1  
f  x2 dx
x2  x0 x2  x1  

• Requiring equally-spaced intervals:

  x  x0  h  x  x0  2h   x  x0  x  x0  2h 
f  x0   f  x1 
x2
I  
x0
 h  2h   h  h 

 x  x0  x  x0  h  
f  x2   dx
 2h  h  

• Integrate and simplify:

ba
I   f  x0   4 f  x1   f  x2 
h
h
12 3 2
10
Quadratic
8 Polynomial
0
3 5 7 9 11 13 15

• If we use a = x0 and b = x2, and

x1 = (b+a)/2
f  x0   4 f  x1   f  x2 
I  b  a 
6
width
average height

• Error for Simpson’s 1/3 rule
Et  
h 5 4 
f    
b  a  4 
5
f   5
90 2880
O(h )
ba
h
2
Integrates a cubic exactly: f 4 

   0

Composite Simpson’s 1/3 Rule
• As with Trapezoidal rule, can use multiple

applications of Simpson’s 1/3 rule.
• Need even number of intervals
– An odd number of points are required.

• Example: 9 points, 4 intervals

25
20
15
10
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
-5

• As in composite trapezoid, break integral up

into n/2 sub-integrals:
I   f  x dx   f  x dx  ...   f  x dx
x 2 x 4 x n
x 0 x 2 x n 2
• Substitute Simpson’s 1/3 rule for each

integral and collect terms.
n 1 n2
f  x0   4  f x  2  f x  f x 
i j n
I  b  a 
i 1,3,5 j  2,4,6
3n
n+1 data points, an odd number
• Odd coefficients receive a weight of 4, even

receive a weight of 2.
• Doesn’t seem very fair, does it?
4 1 coefficients on
1
4 1 i=n numerator
1 1
1 4
i=0

Error Estimate
• The error can be estimated by:
nh5 ( 4 ) b  a h 4 ( 4 ) 4
Ea  f  f O(h )
180 180
• If n is doubled, h  h/2 and Ea  Ea/16
f ( 4 ) is the average 4th derivative

Example
 x2
• Integrate f ( x)  e from a = 0 to b = 2.
• Use Simpson’s 1/3 rule:
ba ab
h 1 x0  a  0 x1  1 x2  b  2
2 2
1
dx  h  f  x0   4 f ( x1 )  f  x2  
2
I  e  x2
0 3
1
  f  0   4 f (1)  f  2  
3
1 0
 (e  4e 1  e 4 )  0.82994
3
Example
• Error estimate: Et  
h 5 4 
f  
90
• Where h = b - a and a <  < b

• Don’t know 
– use average value
5
1  4 1 5f  4
 x0   f  4
 x1   f  4
 x2 
Et  Ea   f 
90 90 3

Another Example
• Let’s look at the polynomial again:

f ( x)  0.2  25x  200 x 2  675x 3  900 x 4  400 x5
– From a = 0 to b = 0.8
ba ab
h  0.4 x0  a  0 x1   0.4 x2  b  0.8
2 2
I   f ( x )dx  h f  x0   4 f ( x1 )  f  x2 
2 1
0 3

(0.4)
 f 0  4 f (0.4)  f 0.8
3
 1.36746667
Exact integral is 1.64053334
Error
• Actual Error: (using the known exact value)

E  1.64053334 - 1.36746667  0.27306666 16%
• Estimate error: (if the exact value is not

available)
h 5 4 
Et   f  
90
• Where a <  < b.

Error
• Compute the fourth-derivative

f ( 4)
( x )  21600  48000 x
0.45 0.45
Et  Ea  f 4 
x1    f 4 
0.4  0.27306667
90 90
middle point
• Matches actual error pretty well.

Example Continued
• If we use 4 segments instead of 1: h
ba
 0.2
– x = [0.0 0.2 0.4 0.6 0.8] n
f (0)  0.2 f (0.2)  1.288 f (0.4)  2.456
f (0.6)  3.464 f (0.8)  0.232
n 1 n 2
f  x0   4  f x   2  f x   f x 
i j n
I  b  a  i 1, 3, 5 j 2, 4,6
3n
f (0)  4 f (0.2)  2 f (0.4)  4 f (0.6)  f (0.8)
 0.8  0
(3)( 4)
0.2  4(1.288  3.464)  2(2.456)  0.232
 0.8
12
 1.6234667 Exact integral is 1.64053334
Error
• Actual Error: (using the known exact value)

E  1.64053334 - 1.6234667  0.01706667 1%
• Estimate error: (if the exact value is not

available)
0.25 4  0.25 4 
Et  Ea   f x2    f 0.4  0.0085
90 90
middle point

Error
• Actual is twice the estimated, why?

• Recall:
f ( 4)
( x )  21600  48000 x
max
x 0,0.8 
 f ( 4)
( x)  f ( 4)
(0)  21600
f ( 4)
(0.4)  2400

Error
• Rather than estimate, we can bound the

absolute value of the error:
0.25  4 0.25  4
Ea   f    f  0   0.0768
90 90
• Five times the actual, but provides a safer

error metric.

Simpon’s 1/3 Rule
• Simpson’s 1/3 rule uses a 2nd order polynomial
– need 3 points or 2 intervals
– This implies we need an even number of intervals.
• What if you don’t have an even number of
intervals? Two choices:
1. Use Simpson’s 1/3 on all the segments except the last
(or first) one, and use trapezoidal rule on the one left.
– Pitfall - larger error on the segment using trapezoid
2. Use Simpson’s 3/8 rule.

• Simpson’s 3/8 rule uses a third order polynomial
– need 3 intervals (4 data points)
f ( x)  p3 ( x)  a0  a1 x  a2 x 2  a3 x 3
I   f  x dx   p3  x dx
x3 x3
x0 x0

• Determine a’s with Lagrange polynomial
• For evenly spaced points
I  h f  x0   3 x1   3 f  x2   f  x3 
3
8
ba
h
3

Error
• Same order as 1/3 Rule.

– More function evaluations.
– Interval width, h, is smaller.
3 5  4
Et   h f   O(h 4 )
80
• Integrates a cubic exactly:

 f 4     0

Comparison
• Simpson’s 1/3 rule and Simpson’s 3/8 rule have
the same order of error
– O(h4)
– trapezoidal rule has an error of O(h2)
• Simpson’s 1/3 rule requires even number of
segments.
• Simpson’s 3/8 rule requires multiples of three
segments.
• Both Simpson’s methods require evenly spaced
data points
Mixing Techniques
• n = 10 points  9 intervals
– First 6 intervals - Simpson’s 1/3
– Last 3 intervals - Simpson’s 3/8
Simpson’s 1/3
Simpson’s 3/8

Newton-Cotes Formulas
• We can examine even higher-order

polynomials.
– Simpson’s 1/3 - 2nd order Lagrange (3 pts)
– Simpson’s 3/8 - 3rd order Lagrange (4 pts)
• Usually do not go higher.
• Use multiple segments.
– But only where needed.

Adaptive Simpson’s Scheme
• Recall Simpson’s 1/3 Rule:

I
h
 f x0   4 f x1   f x2 
3
• Where initially, we have a=x0 and b=x2.

• Subdividing the integral into two:
h
I  f  a   4 f  x1   2 f  x2   4 f  x3   f  b  
6

• We want to keep subdividing, until we

reach a desired error tolerance, .
• Mathematically:
h 
b
 f  x  dx    f  a   4 f  x1   f  b     
a 3 
h 
b
 f  x  dx    f  a   4 f  x1   2 f  x2   4 f  x3   f  b     
a 6 

• This will be satisfied if:
h  
c
 f  x  dx    f  a   4 f  x1   f  x2     , and
a 6  2
h  
b
 f  x  dx    f  x2   4 f  x3   f  b     , where
c 6  2
ab
c  x2 
2
• The left and the right are within one-half of the
error.
• Okay, now we have two separate intervals

to integrate.
• What if one can be solved accurately with
an h=10-3, but the other requires many,
many more intervals, h=10-6?

• Adaptive Simpson’s method provides a

divide and conquer scheme until the
appropriate error is satisfied everywhere.
• Very popular method in practice.
• Problem:
– We do not know the exact value, and hence do
not know the error.

• How do we know whether to continue to

subdivide or terminate?
b
I   f  x  dx  S  a, b   E  a, b  , where
a
ba   ab  
S  a, b    f a  4 f    f  b   , and
6   2  
1  b  a   4
5
E  a, b      f
90  2 

• The first iteration can then be defined as:

I  S    E   , where
1 1
S    S  a, b  , E    E  a , b 
1 1
• Subsequent subdivision can be defined as:

S  2
 S  a, c   S  c, b 

• Now, since
E  2
 E  a, c   E  c, b 
• We can solve for E(2) in terms of E(1).
5 5
1  h / 2   4 1  h / 2   4
E
2
   f    f
90  2  90  2 
5
 1  1  h   4 1 1
 4    f  E
 2  90  2  16

• Finally, using the identity:

1 1  2  2
I S E S E
• We have:
 2 1 1  2  2
S S  E E  15E
• Plugging into our definition:
I S  2
E  2
S  2

15

1  2
S S 1

• Our error criteria is thus:

 2
I S 
15

1  2
S S 1
 
• Simplifying leads to the termination
S   15
formula:  2 1
S

• What happens graphically:

S2  S1  15  subdivide


S2  S1  15  subdivide
2


S2  S1  15  done
4
1
I  S2   S2  S1 
15


S2  S1  15  subdivide
4

1
I  S2   S2  S1 
15

1
I  S2   S2  S1 
15

Iright =Ileft + Iright
I=Ileft + Iright

• We gradually capture the difficult spots.

Adaptive Simpson’s Code
• Simple Recursive Program

static const int m_nMaximum_Divisions = 1000;
Real IntegrationSimpson( const Real (*f) (Real x), const Real start, const Real end, const Real
error_tolerance, int &level )
{
level += 1;
Real h = (end – start);
Real midpoint = (start + end) / 2.0;
Real f_start = f(start);
Real f_end = f(end);
Real f_mid = f(midpoint );
oneLevel = h*( f_start + 4.0*f_mid + f_end) / 6.0;
Real leftMidpoint = (start+ midpoint ) / 2.0;
Real rightMidpoint = (end+ midpoint ) / 2.0
Real f_midLeft = f(leftMidpoint );
Real f_midRight = f(rightMidpoint );
twoLevel = h*(f_start + 4.0* f_midLeft + 2.0* f_mid + 4.0* f_midRight + f_end) / 12.0;
if( level >= m_nMax_Divisions ) // Terminate the process, converging too slow
return twoLevel;

Adaptive Simpson’s Code
if( absf( twoLevel – oneLevel) < 15.0*error_tolerance) // Desired solution reached

return twoLevel + (twoLevel-oneLevel) / 15.0;
//
// Otherwise, split the interval in two and recursively evaluate each half.
//
leftIntegral = IntegrationSimpson( f, start, midpoint , error_tolerance/2.0, level );
rightIntegral = IntegrationSimpson( f, midpoint , end, error_tolerance/2.0, level );
return leftIntegral + rightIntegral;
}

Guassian Quadrature
• Idea is that if we evaluate the function at certain
points, and sum with certain weights, we will get a
more accurate integral
• Evaluation points and weights are pre-computed
and tabulated
• Basic form: I  f ( x)dx   c f ( x )
1 n

1
i 1
i i
ci : weighting factors
xi : sampling points selected optimally New!!

Guassian Quadrature
• Note that the interval is between –1 and 1
• For other intervals, a change of variables is used to
transfer the problem so that it utilizes the interval
[-1, 1]
• This is a linear transform, such that for t[a,b]:
b
a
f (t )dt
• We have for x[-1,1]:

(b  a) x  b  a 2t  b  a
t x
2 ba
Guassian Quadrature
• As t = a  x = -1
• As t = b  x = 1
(b  a )
dt  dx
2
 (b  a) x  b  a 
f (t )  f  
 2 
b (b  a ) 1  (b  a) x  b  a 
 a
f (t )dt 
2 1
f
 2


dx

Guassian Quadrature
• Basic form of Gaussian quadrature:
n
I   f ( x)dx   ci f ( xi )
1
1
i 1
• For n=2, we have:

I  c1 f x1   c2 f x2 
• This leads to 4 unknowns: c1, c2, x1, and x2
– two unknown weights (c1, c2)
– two unknown sampling points (x1, x2)
Guassian Quadrature
• What we need now, are four known values

for the equation.
• If we had these, we could then attempt to
solve for the four unknowns.
• Let’s make it work for polynomials!!!

Guassian Quadrature
• In particular, let’s look at these simple
polynomials:
– Constant
• f(x)=1
– Linear
• f(x)=x
– Quadratic
• f(x)=x2
– Cubic
• f(x)=x3

Guassian Quadrature
• Recalling the formula: I  c1 f x1   c2 f x2 
– Constant
 c1 f  x1   c2 f  x2   c1  c2
1
• f(x)=1  1dx  2
1
 c1 f  x1   c2 f  x2   c1 x1  c2 x2
– Linear 1
• f(x)=x  xdx  0
1
– Quadratic 2
   
1
1     
2 2 2
• f(x)=x2
x dx c1 f x1 c 2 f x 2 c x
1 1 c x
2 2
3
– Cubic
x 3 dx  0  c1 f  x1   c2 f  x2   c1 x1  c2 x2
1

3 3
• f(x)=x3 1

Guassian Quadrature
• We can now solve for our unknowns:

– Note, this is not an easy problem and will not be
covered in this class.
c1  c2  1
1
x1    0.577
3
1
x2   0.577
3

Guassian Quadrature
• This yields the two point Gauss-Legendre

formula
 1   1 
I  f   f  
 3  3

Guassian Quadrature
• This is exact for all polynomials up to and

including degree 3 (cubics).
1 1 1 1 1
  ax  bx  cx  d dx  a  x dx  b  x dx  c  xdx  d  dx
3 2 3 2
1 1 1 1 1
  1 3  1 3    1  2  1  2    1   1  
 a    b  c   d 1  1
  3   3     3   3     3   3  
   
 ax bx cxd 
1
3 2 3

1
3

Guassian Quadrature
1
1
f ( x)dx  f (0.577)  f (0.577)
f(x)
f(-0.577)
f(0.577)
-0.577 0.577
x
-1 1
Example
• Integrate f(x) from a = 0 to b = 0.8
f ( x)  0.2  25x  200 x 2  675x 3  900 x 4  400 x5
• Transform from [0, 0.8] to [-1, 1]
b (b  a ) 1  (b  a ) x  b  a 
a
f (t )dt 
2 1
f
 2


dx
0.8 (0.8  0) 1  (0.8  0)t  0.8  0 

I  f ( x )dx   f dt
0 2 1
 2 
 0.4  f 0.4t  0.4 dt
1
1
Example
• Solving
I  0.4  f 0.4t  0.4 dt
1
1
1 0.2  25(0.4t  0.4)  200(0.4t  0.4)2 

 0.4   5
dt
 675(0.4t  0.4)  900(0.4t  0.4)  400(0.4t  0.4) 
1 3 4
• And substituting for the 2-point formula:

I  0.4 f t dt
1
1
t  1 / 3
I  0.51674055  1.30583723  1.82257778
Higher-order Gaussian Quadrature
• Recall the basic form:

n
I   f ( x)dx   ci f ( xi )
1
1
i 1
• Let’s look at n=3.

I  c1 f x1   c2 f x2   c3 f x3 
• We now have 6 unknowns: c1, c2, c3,x1, x2, and x3
– three unknown weights (c1, c2 , c3)
– three unknown sampling points (x1, x2 , x3)

Use 6 equations - constant, linear, quadratic, cubic,
4th order and 5th order to find those unknowns
 c1 f  x1   c2 f  x2   c3 f  x3   c1  c2  c3
1
 1dx  2
1
 c1 f  x1   c2 f  x2   c3 f  x3   c1 x1  c2 x2  c3 x3
1
 xdx  0
1
2
     
1
1       
2 2 2 2
x dx c1 f x1 c2 f x 2 c3 f x3 c x
1 1 c x
2 2 c x
3 3
3
x 3dx  0  c1 f  x1   c2 f  x2   c3 f  x3   c1 x1  c2 x2  c3 x3
1

3 3 3
1
2
     
1
1       
4 4 4 4
x dx c1 f x1 c2 f x 2 c3 f x3 c x
1 1 c x
2 2 c x
3 3
5
x 5dx  0  c1 f  x1   c2 f  x2   c3 f  x3   c1 x1  c2 x2  c3 x3
1

5 5 5
1

• Can solve these equations (or have some one

smarter than us, like Guass solve them).
c1  5 / 9 c2  8 / 9 c3  5 / 9
x1   3 / 5  0.77459669 x2  0 x2  3 / 5  0.77459669
• Produces the three point Gauss-Legendre formula

I  c1 f x1   c2 f x2   c3 f x3 
– Exact for polynomials up to and including degree 5
(because using 5th degree polynomial)

1 5 8 5
1 f ( x)dx  9 f (0.775)  9 f (0.0)  9 f (0.775)
f(-0.775)
f(0)
f(0.775)
x
-0.775 0.775
February 4, 2020 -1 OSU/CSE 541 1 114
Example
f ( x)  0.2  25x  200 x 2  675x 3  900 x 4  400 x5

Integrate from a = 0 to b = 0.8
Transform from [0, 0.8] to [-1, 1]
f x dx
0.8
I  replace -0.4 with +0.4
0
10.2  25(0.4t  0.4)  200(0.4t  0.4)2 

  5
dt
 675(0.4t  0.4)  900(0.4t  0.4)  400(0.4t  0.4) 
1 3 4

Example
• Using the 3-point Gauss-Legendre formula:
 3 8  3
 f 0 
5 5
I 
f   f  
9 
5 9 9 5 
  
Substitute into the transform equation and get
I  0.281301290  0.873244444  0.485987599

 1.64053334
Gaussian Quadrature
Can develop higher order Gauss-Legendre forms

using
I  c1 f x1   c2 f x2   ...  cn f xn 
Values for c’s and x’s are tabulated

Use the same transformation to map interval onto
[-1, 1]

I   f ( x)dx  c1 f x1   c2 f x2   ...  cn f xn 
1
1
n 2 3 4 5 6
ci 1.0 0.5555555556 0.3478548451 0.2369268850 0.1713245
1.0 0.8888888889 0.6521451549 0.4786286705 0.3607616
0.5555555556 0.6521451549 0.5688888889 0.4679139
0.3478548451 0.4786286705 0.4679139
0.2369268850 0.3607616
0.1713245
 0.5773502692  0.7745966692  0.8611363116  0.9061798459  0.932469514

0.5773502692 0.0000000000  0.3399810436  0.5384693101  0.661209386
xi 0.7745966692 0.3399810436 0.0000000000  0.238619186
0.8611363116 0.5384693101 0.238619186
0.9061798459 0.661209386
0.932469514
Gaussian Quadrature
• Requires function evaluations at non-
uniformly spaced points within the
integration interval
– not appropriate for cases where the function is
unknown
– not suited for dealing with tabulated data that
appear in many engineering problems
• If the function is known, its efficiency can
be a decided advantage
Gaussian Quadrature
• Problems:
– If we add more data points, like doubling the
number of sample points.

Kuliah5.1 - (Integrasi Numerik)

Uploaded by

Copyright:

Available Formats

Kuliah5.1 - (Integrasi Numerik)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Kuliah5.1 - (Integrasi Numerik)

Uploaded by

Copyright:

Available Formats

Numerical Integration

February 4, 2020 OSU/CSE 541 2

Basic definition of an integral: f(x)

sum of height  width x

– Integrand is not precisely defined by an equation, i.e.,

in which wi are the weights, xi are the sampling

February 4, 2020 OSU/CSE 541 5

• Divide [x0 , xn] into n intervals (n1)

February 4, 2020 OSU/CSE 541 6

• Assume that f(x)>0 everywhere.

February 4, 2020 OSU/CSE 541 7

February 4, 2020 OSU/CSE 541 8

• Likewise, still assuming that f(x)>0

February 4, 2020 OSU/CSE 541 9

February 4, 2020 OSU/CSE 541 10

• As n, one would assume that the sum of the

February 4, 2020 OSU/CSE 541 11

February 4, 2020 OSU/CSE 541 12

• Halving each interval (much like Lab1):

February 4, 2020 OSU/CSE 541 13

• One more time:

February 4, 2020 OSU/CSE 541 14

• Note that if a function is monotonically

• Thinking back to lab1, what were the limits

February 4, 2020 OSU/CSE 541 16

February 4, 2020 OSU/CSE 541 17

• Different choices for m’s lead to different

February 4, 2020 OSU/CSE 541 19

I  width  average height

February 4, 2020 OSU/CSE 541 21

• You get exact integration if the function, f, is

February 4, 2020 OSU/CSE 541 22

Use trapezoidal rule:

February 4, 2020 OSU/CSE 541 23

Estimate error: Et   1 f  h 3

Don’t know  - use average value

February 4, 2020 OSU/CSE 541 25

February 4, 2020 OSU/CSE 541 26

• Results in the Composite Trapezoid Formula:

• Think of this as the width times the average

February 4, 2020 OSU/CSE 541 28

• The error can be estimated as:

• Where, f  is the average second derivative.

• Integrate: f x   0.3  20 x  140 x 2  730 x 3  810 x 4  200 x5

• A single application of the Trapezoid rule.

February 4, 2020 OSU/CSE 541 31

• We don’t know  so approximate with

f x   280  4380 x  9720 x 2  4000 x 3

• The error can thus be estimated as:

February 4, 2020 OSU/CSE 541 33

February 4, 2020 OSU/CSE 541 34

• Use intervals (0.2,0.4),(0.4,0.6),(0.6,0.8):

February 4, 2020 OSU/CSE 541 36

• Use intervals (0.2,0.3),(0.3,0,4), etc.

f 0.2   2 f 0.3  f 0.4   f 0.5  f 0.6  f 0.7   f 0.8

True value of integral is 12.82

February 4, 2020 OSU/CSE 541 38

• Consider a two-dimensional case.

• For the Trapezoid Rule, this leads to