Você está na página 1de 16

Solving Mathematical Equations Using

Numerical Analysis Methods

Bisection Method, Fixed Point Iteration,


Newton’s Method
Prepared by
Parag Jain~Mohamed Toure
Dowling College, Oakdale, NY
For Research Topics in Computer Science
---The Computer Science Lyceum---
Spring 2002
CONTENTS

Notice
Abstract
Introduction
Bisection Method
Fixed-Point Iteration Method
Newton’s Method
Conclusion
Acknowledgements
References
NOTICE
This project and the contents herein are provided as a
convenience to the user. The contents of this project are
provided on "as is" and "as available" basis. We have done
this work in good faith. This material is purely for
educational purpose and the user of this document should
have no commercial intention. We are not responsible for
any damage this material might cause.
NO WARRANTY OF ANY KIND IS MADE IN
RELATION TO THE AVAILABILITY, ACCURACY,
RELIABILITY OR CONTENT OF THIS MATERIAL.
Suggestions related to how to improve this material are
welcome. Email us at: Fausto09@yahoo.com or
Jain9164@dowling.edu
ABSTRACT

In our daily life we encounter several problems that need to


be analyzed and solved. In Mathematics, it is usually
impossible to solve to exactitude real world problems. It is
therefore useful to have an idea about what the solution to
the problem would look like.
We, have been working in Numerical Analysis since long,
and the challenge involved motivated us to implement a
nice GUI for the user to help him solve problems using
certain methods of approximation such as:
Bisection Method
Fixed Point Iteration Method
Newton’s Method
INTRODUCTION

Our program aims at finding the roots of mathematical equations


utilizing the three methods – Bisection Method, Fixed Point Iteration
Method, and Newton’s Method.
The idea is to use a library of equations from which the user chooses
one, selects the method that he/she wants to utilize for solving that
equation, defines an interval or initial guess (as the case might be), and
defines the accuracy.
The program then finds the solution to the equation to the defined
accuracy (say 1*10-10) and to achieve that performs iterations of the
chosen method.
The program displays the roots for the equation and the number of
iterations performed.
INTRODUCTION CONTD…

A slightly different approach is used while solving an equation using


Fixed Point Iteration and Newton’s Method. This approach is called
‘Sampling’.
We utilize the Bisection Method to determine those regions in the
interval where the function has a root, and we look for the root in these
semi-intervals only using Fixed Point Iteration or Newton’s Method.
This expedites the root finding process and reduces the probability of
failure of these two methods.
The failure happens when we start looking for roots in an interval that
is not continuous, has no roots, function not self-mapped (in Fixed-
Point Iteration Method), or the initial guess leads to a critical point (in
Newton’s Method).
BISECTION METHOD

The Bisection Method gives a proof of the Intermediate Value


Theorem and provides a practical method to find roots of equations at
the same time.
Recalling the Intermediate Value Theorem, we have that if f(x) be a
continuous function on the interval [a,b] and if d belongs to [f(a),f(b)],
then there is a point c in [a,b] such that f(c) = d.
By replacing f(x) by [f(x) – d], we may assume that d = 0; it then
suffices to obtain the following version as a rule for the Bisection
Method - let f(x) be a continuous function on the interval [a,b]. If f(a)
and f(b) have opposite signs, then there is a point c in [a,b] such that
f(c) = 0.
BISECTION METHOD CONTD…
We illustrate Bisection Method by considering the following
polynomial p(x) = x7 + 9x5 - 13x - 17
Note that p(0)=-17 and p(2)=373. Therefore, since p(x) is a continuous
function (i.e., its graph has no “breaks”), we know that there must be a
root, say r, in the interval (0,2) by the Intermediate Value Theorem.
To shrink the interval in on r, we now evaluate p at the midpoint of
(0,2) which is 1. p(1)=-20. Now we see that r must actually lie in the
interval (1,2) since p switches signs from negative to positive as x
ranges from 1 to 2. So we have reduced the interval under
consideration from (0,2) to (1,2). We have cut the length of our
interval in half, or bisected it.
We look next at the midpoint of (1, 2), namely 1.5, and p(1.5)=48.929.
Thus, r must be in the interval (1, 1.5). We continue this procedure
until a desired accuracy has been achieved. The table summarizes the
results of iterating this technique several times.
A plot of these values follows a general pattern shown in the graph.
This is how our applet works.
FIXED POINT ITERATION METHOD
This method is based on a different principle than that of bisection. In
this case, one seeks the roots of the equation x = g(x).
By definition, a fixed point for a function g(x) is the number x that
verifies g(x) = x. Note that if one defines f(x) = x - g(x) = 0 one is
reduced to seeking the zero of the function f(x) = 0. One gets an idea of
the zero’s location by using the Bisection Method through ‘Sampling’
as discussed earlier.
This method works only if g(x) is continuous and self-mapped.
We assume that we are seeking the zeros of f(x) = 0. This problem can
be transformed to the fixed point g(x) = x form in several ways:
g(x) = f(x) + x = x
g(x) = h(x)f(x) + x = x (h(x) != 0 in the interval of x we are considering the
solution to lie in.)
When solving the equation x = g(x) we start with an initial guess and
perform iterations with subsequent values of g(x) for x.
Example
NEWTON’S METHOD
Newton’s Method is a very efficient way to approximate
the roots of equations of the type f(x) = 0, where f is a
given function.
Suppose that you start at a point x0 you think is close to the
root. The tangent line at the point x0 is pretty close to the
curve. Hence, the intersection point of the tangent line and
the x-axis should be closer to the root.
Since the slope of the tangent line is f’(x0), that point of
intersection is x1 = x0 – f(x0)/f’(x0). So x1 is a better
approximation than x0
If the result is not up to the expected approximation then
the same iteration is used with x1 as the initial guess to
generate x2
NEWTON’S METHOD CONTD…

As you have noticed, Newton’s Method is based on iteration and


uses f’(x) in addition to f(x).
It also gives a fixed-point function g(x) given f(x) as follow:
xn = xn-1 - f(xn-1)/f’(xn-1) where n=1, 2, 3, …
x0 is the starting point
xn = g(xn-1)
The previous algorithm is used to generate a sequence of points
{xn}n=0 that converges to x, where x is the true, exact root of f.
For a lot of functions, this iterative method works very well. But not
for all functions. For instance, it doesn't work for any function that
doesn't have a root. How could it? And it doesn't always work for
some functions that do have roots, for instance, the cube root
function.
Let’s examine the following graphs
NEWTON’S METHOD CONTD…

How can one implement such a nice algorithm bounded by


chances of failure?
The final approach that we have adopted is to use an already
defined method that confirmed the existence of a root.
The idea is to use the Bisection Method in order to sample
the values of the given function, and determine whether or
not there is a change in sign. If yes, we use Newton’s
Method with the initial guess at a point chosen in that
interval that generated the change in sign.
This ensures that there is a root, and this process will
accelerate Newton’s Method.
NEWTON’S METHOD CONTD…
However, there might be more than one solution to the problem. The next
approach will be to divide the initial function by (x-previous root), and
regenerate the process previously described.
If the solution goes to the same root as the previous, then that root has an
order of multiplicity.
Repeat the process until there is no more root (i.e. f(x) = constant or an
equation with imaginary roots).
It is easy to show that Newton’s Method fails infinitely many times
depending on the point chosen as the initial guess.
The sampling actually avoids those fatalities (i.e. it reduces the probability
of failure by choosing the initial point as close as possible to a critical
point of the function).
If this algorithm works, it is one of the most efficient in approximation
theory.
CONCLUSION

The project demonstrated how we could find roots of equations using


suitable techniques.
The heuristic that we implement in the program for Fixed-Point
Iteration and Newton’s Method makes these methods faster and
workable even in situations in which they would have otherwise
failed.
Having a predefined library of equations limits the applicability of the
program so we are still working on the implementation of a parser in
our project.
We are still in a process of exploring how we could make our root-
finding methods better, more user-friendly, and more efficient. It is an
unending quest.
ACKNOWLEDGEMENTS

We acknowledge the support and encouragement that we


got from Professor Herbert J. Bernstein throughout this
semester. He was a pathfinder to us and his guidance was
really invaluable.
Also we acknowledge Professor Raymond Grinnell for the
masterminded mathematical help that we got from him.
Thanks to all our colleagues and friends for their
camaraderie.
REFERENCES
Richard L. Burden & J. Douglas Faires, Numerical Analysis (Boston, MA:
PWS Publishing Company, 1993).
http://archives.math.utk.edu/topics/
http://perso.wanadoo.fr/jean-pierre.moreau/index.html
http://www.ifi.uio.no/~pde/examples/doc/AllNames.html
http://members.tripod.com/~showing/VisualMath.html
http://www.csulb.edu/~wziemer/FixedPoint/FixedPoint.html
http://www.dgp.toronto.edu/people/JamesStewart/applets/roots/roots.html
http://delphi.about.com/library/bluc/blrtlmain.htm?
iam=dpile&terms=oepn+source+for+math+function
http://donald.phast.umass.edu/kicons/greek.html
http://mss.math.vanderbilt.edu/~pscrooke/toolkit.html
http://www.cs.colorado.edu/~main/ds/daybyday.html

Você também pode gostar