|
MATHEMATICS - THEORY
|
|
|
In some situations, it is impossible to find the exact
value of a definite integral. For these cases, an approximate value of
definite integrals are needed. In this section, several rules to
find the approximate value of a definite integral are introduced.
|
|
|
|
|
|
Midpoint Rule
|
Midpoint Rule
|
|
Since the
definite integral is defined as a limit of Riemann sum, any Riemann sum could be used as an approximation to the integral. Recall, a definite integral is
where f(x) is a continuous function on interval [a, b]. To apply an approximation, first divide the interval [a, b] into n subintervals of equal length
Δx = (b - a)/n
Next, choose a x in each subinterval and the integration can be approximated as
where xi* is any point in the ith subinterval
[xi-1,xi].
If xi* is choose to be the midpoint of each
subinterval, this approximation is called the Midpoint Rule.
xi* = 1/2(xi +
xi+1)
The definite integral of function f(x) in an interval [a, b] equals
the area bounded by curve y = f(x), x = a, x = y, and y = 0. Therefore,
Midpoint Rule actually is using a rectangle to approximate the area under
the curve over a subinterval [xi-1 , xi]. |
|
|
|
|
|
Trapezoidal Rule
|
Trapezoidal Rule
|
|
When using the Riemann sum to approximate definite
integration, the approximation is called left endpoint approximation
if the left point is choose as xi*. In the same
way, the approximation
is called right endpoint approximation if the right point is choose as
xi* . The Trapezoidal Rule is from
averaging these two approximation.
Actually the Trapezoidal Rule is using a trapezoid
to approximate the area over a subinterval. For example, in the interval
[x0,
x1], the area of the trapezoid is
ΔA = Δx/2[f(x0) + f(x1)]
When all the trapezoids at each interval are added together, the total
gives,
A = Δx/2[(f(x0) + f(x1))
+ (f(x1) + f(x2) + ...
+ (f(xn-1) +
f(xn))]
=
Δx/2 + [f(x0) + 2f(x1)
+ ... + 2f(xn-1) + f(xn)]
which is the Trapezoidal Rule approximation.
|
|
|
|
|
|
Simpson's Rule
|
Simpson's Rule
Simpson's Rule on Interval [-h, h] |
|
Simpson's Rule for approximation integration
uses parabolas to approximate the curved edge. First divide interval
[a, b] into n equal length subintervals
h = Δx = (b - a)/n
where n is an even number.
On each consecutive pair of intervals, the curve y = f(x) can be approximated
by a parabola y = Ax2 + Bx + C. To better understand this, consider
a subinterval between x0 and x2. To simply the calculation,
assume
x0 = -h
x1 = x0 + h = 0
x2 = x0 + 2h = h
Then the area under the parabola is
Since point P0, P1 , and P2 are on
both curves,
y0 = f(x0) = A(-h)2 - Bh + C
y1 = f(x1) = C
y2 = f(x2) = A(h)2 + Bh + C
Adding the above three equations together gives
y0 + y1 + y2 = 2Ah2 + 3C
Therefore, the area under the parabola can be rewritten as
A = h/3( 2Ah2 + 6C)
= h/3(y0 + y1 +
y2 + 3y1)
= h/3(y0 + 4y1 + y2)
Shifting the parabola between [-h, h] to [x0,
x2] does not change the area under it. Similarly, the area
under the parabolas through point P2, P3, and
P4 are
A = h/3(y2 + 4y3 + y4)
|
|
|
|
Simpson's Rule on
Interval [a, b] (n = 6)
|
|
Calculating the area under all the parabolas and adding them together gives
If n = 6, the integration can be approximated as
Note the pattern of coefficients is: 1, 2, 4, 2, 4, 2, ..., 4, 2,
4, 1. |
|
|
|
|