MATLAB for MAPH 3071 Lab 6
Numerical Integration
Numerical integration is used where an integral is impossible or difficult to solve
analytically.
1. Composite Trapezoidal Rule
The approach is to subdivide the interval across which we wish to integrate and
approximate the integral of the function with a number of trapezoids. Each
subdivision is of equal width h, where
h = (b-a)/n where n is the number of subdivisions.
∫
h
[ f ( a) + 2 f ( a + h) + 2 f ( a + 2 h) + 2 f ( a + 3h ) + ........+ 2 f (a + ( n − 1)h ) + f (b )]
b
f ( x ) dx ≈
a 2
Here is a program on the class library (L:\trapex.m) which integrates exp(-x2) over a
given interval using the composite trapezoidal rule.
L:\trapex.m
% trapex.m test program for numerical integration using the composite
% trapezoidal rule to solve the integral of exp(-x^2) between
% a and b
clear; help trapex;
format long; % configures MATLAB to report numbers with more decimal places
a=input('input a (starting value)->');
b=input('input b (end value) ->');
n=input('input number of intervals (n) ->');
h=(b-a)/n;
fa=exp(-a^2); % f(a)
fb=exp(-b^2); %f(b)
ff=0;
for i=2:n
ff=ff+(2*exp(-(a+(h*(i-1)))^2)); % sum of 2f(a + i(h)) where i = 1 to n-1
end
result=(h/2)*(fa+fb+ff)
% result = f(a) + f(b) + sum of 2f(a + i(h)) where i = 1 to n-1
1
2. Composite Simpson’s Rule
The approach is similar to the trapezoidal rule, however with Simpson’s rule we
subdivide the interval across which we wish to integrate and approximate the integral
of the function with a number of quadradics
h f (a ) + 4 f (a + h) + 2 f (a + 2h) + 4 f (a + 3h) + 2 f (a + 4h ) + .....
∫
b
f ( x )dx ≈
3 ...................... + 2 f (a + (n − 2) H ) + 4 f ( a + (n − 1)h) + f (b)
a
As with the composite trapezoidal rule the subintervals must have equal width, and n
must be an even number.
Note the alternating coefficient of the middle terms – it goes 4,2,4,2 and so on. We
can achieve this in MATLAB using a for loop. For example, if we use n=10, we want
the following
h/3 * [ f(a) + 4 f (a+h) + 2 f (a+2h) + 4 f (a+3h) + 2 f (a+4h) + 4 f (a+5h) +
2 f(a+6h) + 4 f (a+7h) + 2 f (a+8h) + 4 f (a+9h) + f (b) ]
One way to achieve this in MATLAB is to split the middle terms into two parts as
follows;
Step (1)
» n=10;
» for i=2:2:10
h=i-1
end
Result of this for loop => h = 1,3,4,5,7,9 which are the 'n' coefficients we require for
the
4 f(a+nh) terms.
Step (2)
» for i=3:2:n
h=i-1
end
Result of this for loop => h = 2,4,6,8 which are the 'n' coefficients we require for the
2 f(a+nh)
Here is a program on the class library (L:\simpsonex.m) which integrates exp(-x2)
over a given interval using the composite simpson’s rule.
2
L:\simpsonex.m
% Simpsonex.m - A program for composite Simpsons 1/3 rule
% to numerically integrate a function between a and b with
% n subintervals.
% The example program integrates exp(-x^2)
format long % sets MATLAB to report more decimal places
clear; help simpsonex;
a=input('input a (starting value)->');
b=input('input b (end value) ->');
n=input('input number of intervals (n) ->');
h=(b-a)/n; % interval with
fa=exp(-(a^2)); % f(a)
fb=exp(-(b^2)); % f(b)
ff=0;
for i=2:2:n; % all 4*f(a+nh) terms to f(b) h=(1,3,5,7,…..,n-1)
x = (a+(i-1)*h);
fx = exp(-x^2);
ff = ff + 4*fx;
end
for i=3:2:n; % all 2*f(a+nh) terms to f(b) h=(2,3,4,6,…..,n-2)
x = (a+(i-1)*h);
fx = exp(-x^2);
ff = ff + 2*fx;
end
result=(h/3)*(fa+fb+ff) % integral result
% with approximation to area under curve
END
