Scribd
Upload
23 views

Appendix 2: The Solution of Non-Linear Equations

This document discusses methods for finding solutions to non-linear equations. It describes three principal methods: the bisection method, secant method, and Newton-Raphson method. The bisection method uses successive bisections of intervals to narrow in on a solution. The secant method finds solutions by computing the intersection of secant lines. The Newton-Raphson method iteratively finds better approximations using the derivative to converge rapidly on solutions. An example applying all three methods to solve an equation is provided and illustrates the convergence rates of each approach.

Uploaded by

akita_1610
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
23 views

Appendix 2: The Solution of Non-Linear Equations

This document discusses methods for finding solutions to non-linear equations. It describes three principal methods: the bisection method, secant method, and Newton-Raphson method. The bisection method uses successive bisections of intervals to narrow in on a solution. The secant method finds solutions by computing the intersection of secant lines. The Newton-Raphson method iteratively finds better approximations using the derivative to converge rapidly on solutions. An example applying all three methods to solve an equation is provided and illustrates the convergence rates of each approach.

Uploaded by

akita_1610
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Appendix 2: The Solution of Non-linear

Equations

The problem of nding the solution (or solutions) of a non-linear equation


occurs in many branches of mathematics. In compound interest problems,
such an equation is often in a polynomial form (but this is not necessarily
always the case).
Simple examples of non-linear equations are

CONTENTS
Bisection
Method .............314

6x  3x  1 0

(A.2.1)

x  ln x  2 0

(A.2.2)

Secant Method
(Also Known as the
Modified Regula
Falsi Method)...314

ex  sin x 0

(A.2.3)

NewtoneRaphson
Method .............315

Such an equation may be considered abstractly in its general form as


f x 0

Exercises ..........317
Solutions...........318

Although we may not be able to determine the exact value of a solution of the
equation (i.e., the precise value of x such that f (x) 0), in practice we may be
able to estimate the value of x to any desired degree of accuracy. This means
that, given any very small positive number (such as 10e8), we can nd
a number h such that j x e h j < . For practical purposes, h is then regarded as
a solution of the equation.
Special techniques exist for solving polynomial equations; however, the
methods which are most suitable in the more general case can be applied easily
to polynomial equations. Accordingly, we restrict our attention in this
appendix to three of the principal methods which may be applied to solve the
general non-linear equation.
Before one tries to nd the root (or roots) of any equation, it is important to
conrm by as elementary methods as possible that the equation does, indeed,
have at least one root, thereby avoiding wasted time. In addition, one should
try to determine the number of roots and nd some information about their
location.
In almost all practical problems, the function f will be continuous, and
henceforth we shall assume that this is so. (If f is not continuous, the equation

313

314

Appendix 2: The Solution of Non-linear Equations

f (x) 0 may require more careful analysis.) If a < b and f (a) is of opposite sign
to f (b), there exists x with a < x < b such that f(x) 0. This well-known
property can be used in most cases to nd an interval containing a solution of
the equation and leads naturally to one method for nding it.

BISECTION METHOD
Suppose that a0 < b0 and f(a0) f (b0) < 0. Put m (a0 + b0)/2. Then either
(a) f (a0) f (m)  0, in which case there is a solution in the interval [a0, m]; or
(b) f (a0) f (m) > 0, in which case f (m) f (b0) < 0 so that there is a solution in
the interval [m, b0].
We have therefore determined an interval, [a1, b1] say, contained in the interval
[a0, b0] and one half its length, in which the solution lies. Repeating this
procedure, we may construct a series of nested intervals [ai, bi], each containing
the root, such that (bi e ai) (b0 e a0)/2i. By making i sufciently large, we can
therefore nd the solution to any desired degree of accuracy.
The principal advantage of the bisection method lies in the fact that, once an
interval has been found which contains a solution, then the method guarantees
that, provided we are willing to carry out a sufcient number of iterations, the
value of the solution may be found to any specied degree of accuracy. The
major drawback is that, if very high accuracy is required, the necessary number
of iterations may be large. With a modern computer, this may not be a serious
objection, but it is natural to seek faster methods.
In the following sections, we describe two further methods, both of which are
generally satisfactory in nearly all practical situations.

SECANT METHOD (ALSO KNOWN AS THE MODIFIED


REGULA FALSI METHOD)
A sequence {x0, x1, x2, .} is constructed in such a way that as n increases xn
tends to the desired solution. The values of x0 and x1 are chosen arbitrarily, but
in the neighbourhood of the solution. For n  1, xn+1 is dened to be the
x-coordinate of the point of intersection of the x-axis and the line joining the
points xn1 ; f xn1  and xn ; f xn  (see Figure A.2.1).
It is easily veried that
xn1 xn  f xn

xn  xn1
f xn  f xn1

f xn xn1  f xn1 xn
f xn  f xn1

NewtoneRaphson Method

FIGURE A.2.1
Secant method

Given > 0, we continue the iteration until


f xn  f xn  0
at which stage we know that xn is within of the solution.

NEWTONeRAPHSON METHOD
Finally, we consider a technique which gives theoretically faster convergence to
a solution than either of the two preceding methods, but which requires that we
are able to evaluate the derivative of f.
We try to construct a sequence {xn} which converges rapidly to the required
solution. Intuitively, the rationale for the method may be described as follows.
Suppose that xn is close to the solution so that j f (xn) j is small. We wish (xn + h)
to be a better approximation. If h is small,
f xn h x f xn hf 0 xn

315

316

Appendix 2: The Solution of Non-linear Equations

Hence, if
h f xn =f 0 xn
it follows that f(xn + h)x0. Accordingly, we let
xn1 xn  f xn =f 0 xn
and hope that the sequence dened (from a given starting point x0) converges.
If it does, the limit point x is a solution of the equation. The method is illustrated in Figure A.2.2.
The principal advantage of this method lies in its speed of convergence;
however, the choice of the initial point x0 may be of critical importance in the
search for a particular solution. A bad starting value may lead to a solution
other than that sought or even to a non-convergent sequence. Also, the
method does not produce a small interval containing the solution. Usually,
the iterative procedure is continued until jxn1  xn j is sufciently small. In
practice, this is generally an acceptable criterion, but is not entirely foolproof

FIGURE A.2.2
NewtoneRaphson method

Exercises

EXAMPLE A.2.1
Find the positive solution of Eq. A.2.1 using the three
methods discussed.

Solution
Here we have
f x 6x 4  3x  1
and

f (x) is positive for large positive and large negative values


of x, it follows that the equation has two real solutions. By
our earlier remarks, we see that one solution lies in the
interval (e1, 0) and the other in the interval (1/2, 1). For the
bisection method, we put a0 0.5, b0 1; for the secant
method, we let x0 0.5 and x1 1; for the NewtoneRaphson
method, we let x0 0.75.
The number of iterations required by each method depends
on the accuracy to which the solution is required. Our results
are summarized in Table A.2.1. The speed of convergence of
the three methods should be noted.



1
f 0 x 24 x 3 
8
Hence, f is a decreasing function for x < 1/2 and an
increasing function for x > 1/2. Also f (1/2) e 17/8. Since

Table A.2.1 Solution of 6x4 e 3x e 1 0


Accuracy*
[ 10

L4

[ 10L8

Method of Solution

Solution

Number of Iterations

Solution

Number of Iterations

Bisection
Secant
NewtoneRaphson

0.8831
0.8831
0.8831

12
5
3

0.883 111 66
0.883 111 66
0.883 111 67

25
7
4

*The true root lies within of the value given.

and, theoretically, an additional test (such as that described previously for the
secant method) should be applied.
Although the calculation of the derivative may be somewhat laborious and
extra care may be needed initially, the NewtoneRaphson method may produce
signicant savings in time when many similar calculations are required.

EXERCISES
A.1 Find the negative solution of Eq. A.2.1.
A.2 Find all the real solutions of Eq. A.2.2.
A.3 How many real solutions does equation A.2.3 have? Find the two smallest
solutions.

317

318

Appendix 2: The Solution of Non-linear Equations

SOLUTIONS
(To ve signicant gures.)
A.1 0.31391.
A.2 0.15859, 3.1462.
A.3 An innite number. 0.58853, 3.0964.

You might also like