Numerical Integration with Simpson's Rule
Instead of rectangles and trapezoids Simpson's rule uses parabolic arcs. The formula obtained for Simpson's rule requires using uniform width subintervals and an even number of them.
Consider
on
and say we want to find the area. We could use rectangles as in a Riemann sum; however many graphs have curvature as so maybe using strips with parabolic arc will yield better estimates with less work.
That is instead if rectangles or trapezoids:
![numerical integration with the simpson rule _gr_3.gif]](pages/numerical-integration-with-the-simpson-rule/Images/numerical-integration-with-the-simpson-rule_gr_3.gif)
let's use "strips" with parabolic arcs
![numerical integration with the simpson rule _gr_4.gif]](pages/numerical-integration-with-the-simpson-rule/Images/numerical-integration-with-the-simpson-rule_gr_4.gif)
If
is a partition of
with
and
and if we pass a parabolic arc through the points, three at a time say, the points with
-coordinates
then those with
and so on. It can be shown that the area of the region under the parabolic curve
on the interval
has area given by
![]()
where
Thus, to estimate
we can add up the strips as follows:
Definition (Simpson's Rule) The Simpson rule estimates the definite integral of
over
using the formula,
![numerical integration with the simpson rule _gr_19.gif]](pages/numerical-integration-with-the-simpson-rule/Images/numerical-integration-with-the-simpson-rule_gr_19.gif)
where
and
is an even integer.
Example (Simpson's Rule) Consider
on
and let
be a partition of the interval
(so
).
![]()
![numerical integration with the simpson rule _gr_29.gif]](pages/numerical-integration-with-the-simpson-rule/Images/numerical-integration-with-the-simpson-rule_gr_29.gif)
which is a much better estimation than using rectangles with the same partition.
Definition (Error in the Simpson Rule) If
has a continuous fourth derivative on
then the error
(n even) in approximating
by the Simpson rule satisfies:
![]()
where
is the maximum value of
on
Definition (Error in the Simpson Rule) Consider
on
and let
be a partition of the interval
. Find the error in the approximation
![]()
Solution. We compute
,
and
.
We plot the fourth derivative obtaining:
![numerical integration with the simpson rule _gr_47.gif]](pages/numerical-integration-with-the-simpson-rule/Images/numerical-integration-with-the-simpson-rule_gr_47.gif)
and so there is a maxmium at
We obtain
Therefore, the error is less than
![numerical integration with the simpson rule _gr_59.gif]](pages/numerical-integration-with-the-simpson-rule/Images/numerical-integration-with-the-simpson-rule_gr_59.gif)
![numerical integration with the simpson rule _gr_69.gif]](pages/numerical-integration-with-the-simpson-rule/Images/numerical-integration-with-the-simpson-rule_gr_69.gif)
Cite this as:Numerical Integration With The Simpson Rule
Published by Library of Math -- Online math organized by subject into topics.
Written by Smith, David A.
http://www.libraryofmath.com/numerical-integration-with-the-simpson-rule.html


