Professional Documents
Culture Documents
9.2
9.3
9.4
9.5
Regula-Falsi Method
New ton-Raphson Method
Convergence Criterion
Some Results on Roots of Polyno~nialEquations
9.6 Birge-Vieta Method
9.7 Homer'sMethod
9.8 Muller's Method
9.9 Graeffe's Root Squaring Method
9.10 Summary
9.11 Answers to SAQs
9.1 INTRODUCTION
In the last unit we discussed two iteration methods for finding roots of an equation
f (x) = 0. There we have shown that a root of the equationf (x) = 0 can be obtained by
writing the equation in the form x = g (x). Using this form we generate a sequence of
approximations xi + 1 = g (xi) for i = 0,1,2, . . . . In this unit, we shall first discuss two
Objectives
After studying this unit you should be able to :
apply regula-falsi and secant methods for finding the roots,
Solution of Equatiorrci
9.2
The Latin word 'Regula Falsi' means rule of falsehood. It does not mean that the rule is a
false statement. But it conveys that the roots that we get according to the rule are
approximate roots and not necessarily exact roots. The method is also known as the
method of false position. This method is similar to the bisection method you have learnt
in Unit 8.
The bisection method for finding approximate roots bas a drawback that it makes use of
only the signs off (a) andf (6). It does not use the values off (a) ,f (b) in the
computations. For example, iff (a) = 700 andf (b) = - 0.1, then by the bisection method
the first approximate value of a root off (x) is the mid value xo of the interval ]a, b[. But
at xo,f (xo) is nowhere near 0.Therefore in this case it makes more sense to take a value
near to - 0.1 than the middle value as the approximation to the root. This drawback is to
some extent overcome by the regula-falsi method. We shall first describe the method
geometrically.
Suppose we want to find a root of the equation f (x) = 0 where f (x) is a continuous
function. As in the bisection method, we first find an interval [a, b] such that
f (a)f (b) < 0. Let us look at thegraph off (x) given in Figure 9.1.
(a. f(a) )
The conditionf (a) f (b) c 0 means that the point (a,f (a)) and (b,f (b))lie on the opposite
sides of the x-axis. Let us consider the line joining (a, f (a)) and (b,f (b)). This line
crosses the x-axis at some point (c, 0) (see Figure 9.1). Then we take the x-coordinate of
that point as the first approximation. Iff (c) = 0, then x = c is the required root. If
f (a)f (c) < 0, then the root lies in ]a, c[ (see Figure 9.1 (a)): In this case the graph of
y = f (x) is concave near the root r. Otherwise, iff (a) f (c) > 0, the root lies in [s, d ]
(see Figure 9.1 (b)). In this case the graph of y = f (x) is convex near the root. Having
fixed the interval in which the root lies, we repeat the above procedure.
Let us now write the above procedure in the maQematical form. Recall the formda for
the line joining two points in the Cartesian plane. The line joining (a,f (a)) and (b,f (b)) is
given by
Since the straight line intersects the x-axis at (c, O), the point (c, 0) lies on the straight
line. Putting x = c, y = 0 in Equation (9.1), we get
c- a - f o
b-a
b-a
Sdotion,dAlgebraic Plwl
Tr~naeedcoelEqusfjO~~
- f(b)-f(a)
Thus
This expression for c gives an approximate value of a root off (x). Simplifying (9.2), we
can also write it as
Now, examine the sign off (c) and decide in which interval ]a, c[ or ]c, b[, the root lies.
We thus obtain a new interval such that f (x) is of opposite signs at the end points of this
interval. By repeating this process, we get a sequence of intervals ]a, b[,
la, al[, la, a2[, . . . as shown in Figure 9.2.
.A
(b. f(b) )
Figwe 9.2
-'lf
f 01) -f ( a )
Step 3 :Iff (x2) = 0 then x2 is the required root. Iff (x2) # 0 and f ( a )f (x2) < 0, then
the next approximation lies in h o , q[.Otherwise it lies in 1x2, xl[.
Step 4 :Repeat the process till the magnitude of the difference between two successive
iterated values Xi and xi + 1 is less than the accuracy required (note that I
the error after ith iteration).
Xi + 1 - xl
gives
It is known that the equation x? + 7x2 + 9 = 0 has a root between -8 and -7. Use
the regula-falsi method to obtain the root rounded off to 3 decimal places. Stop the
iteration when 1 Xi + 1 - xl 1 <
Solution
Since we are given that xo = - 8 and xl = - 7, we do not have to use step 1. Now
to get the first approximation, we apply the formula in Step 2.
Since,f (.%) = f (- 8) = - 55 andf ( x i ) = f (- 7 ) = Y we obtain
Now we compare the sign off (x2) with the signs off (xo) andf ( X I ) We
. can see
that f (xo) and f (x2) are of opposite signs. Therefore a root lies in the interval
1- 8, - 7.1306t We apply the formula again by renaming the end points of the
interval as x l = - 8 , x2 = - 7.1406. Thcn we get the second approximation as
We repeat this process using steps 2 and 3 given above. The iterated values are
given in the following table.
Table 9.1
Number of
Iterations
Interval
The
Function
Valuef (xi)
From the table, we see that the absolute value of the difference between the 5l.h and
6th iterated values is 1 7.1748226- 7 1747855) = 0.0000371. Therefore we stop
the iteration here. Further, the values off (x) at 6th iterated value is
which is close to zero. Hence we conclude that
0.00046978 = 4.6978 x
-7.175 is an approximate root of x3+ 7x2 + 9 = 0 rounded off to three decimal
places.
It can be notlced that in regula-fa'aisi method, at each stage we find an interval ]xo,xl[
which contains a root and then aFpli iteration formula (9.3). This procedure has a
disadvantage. XI overcome this, regula-falsi method is modified. The modified method is
known as secant method. In this method we choose .Q and xl as any two approximations
of the root. The interval ]xo,x i [ need not contain the root. Then we apply formula (9.3)
with xo , x l , f (XO)andf (XI>.
Note :Geometrically, in secant method, we replace the graph off (x) in the interval
]xn, xn+l[byastraightlinejoiningtwopoints(*., f(*.+]),
f(x,+l))onthe
(r,+I),
curve and take the point of intersectio~iwith x-axis as the approximate value of the root.
Any line joining two points on the curve is called a secant line. That is why this method is
known as secant method (Figure 9.3).
Figure 9.3
(This example was considered in the book 'Numerical methods for scientific and
engineering computation' by M. K. Jain, S. R. K. Iyenger and R. K. Jain.)
Solution
(i)
Therefore the first iterated value is 0.3146653378. To get the 2nd iterated value,
we apply Formula (9.4) with xl = 1, x;! = 0.3146653378. Now
f (1) = - 2.177979523 andf (0.3146653378) = 0.519871175.
Therefore
We continue this process. The iterated values are tabulated in the following
table.
Table 9.2 :Secant Method
Number of
Iterations
Iterated Values xi
f (xi)
p
p
0.3146653378
0.5 19871
0.446728 1466
0.203545
0.53 17058606
- 0.0429311
0.5 169044676
.00259276
0.5 177474653
0.00003011
0.5 177573708
-0.215132~
0.5 177573637
0.178663 x 10 -I2
0.5 177573637
0.222045 x lo-''
From the table we find that the iterated values for 7th and 8th iterations are the
same. Also the value of the function at the 8th iteration is close to zero.
Therefore we conclude that 0.5177573637 is an approximate root of the
equation.
(ii) To apply regula-falsi method, let us first note that f (0) f (1) < 0. Therefore a
root lies in the interval 10, l[. Now we apply Formula (9.3) with
Q = 0 and xl = 1. Then the first approximation is
You may have noticed that we have already calculated the expression on the
right hand side of the above equation in part (i).
Now f (x2) = 0.51987 > 0. This shows that the root lies in the interval
]0.3146653378,1[. To get the second approximation, we compute
which is same as x3 obtained in (i). We findf (x2) = 0.203545 > 0. Hence the
root lies in ]0.4467281446,1[. To get the third approximation, we calculate
The above expression on the right hand side is different from the expression for
xq in part (i). This is because when we use regula-falsi method, at each stage,
we have to check the conditionf (xi) f (xi - < 0.
The computed values of the rest of the approximations are given in Table 9.3.
Table 9.3 :Regula-Falsi Method
No.
Interval
Iterated
Values X i
f (xi)
10, I[
0.3146653378
0.5 19871
10.4467281446, 1[
0.4467281446
0.203545
10.4940153366, 1 [
0.4940153366
10.5099461404, 1[
0.509946 1404
-'
0.236077 x 10 -'
10.5152010099, 1[
0.5152010099
0.776011 x
[0.5176683450, 1[
0.5177478783
0.288554 x 10
]0.5177478783,1[
0.5 177573636
0.396288 x 10
0.708023 x 10
di
t.10g~01- 1.2 = 0
J I ~ L1.
!
(1,)
(')
Fin11 hy I-c~ula-l:~lsi
nicthc?tln root ol 111; cqualio~lx'
betwccll 0.5 and 1.O.
(d)
-+-
.A- - -
1 =- 0, lyiiip
(11)
This method is one of the most useful methods for finding roots of an algebraic equation.
is due to mathematician Isaac Newton. But the method as now used is due to the
mathematician Raphson.
Let us begin with an equation f ( x ) = 0 where f ( x ) and f 'l ( x ) are continuous. Let % be an
initial approximation and assume that xo is close to the exact root a and f' (a)# 0. Let
a = a + h where h is a small quantity in magnitude. Hence f ( a ) = f (xo + h) = 0.
Now we expand f (a+ h ) using Taylor's theorem. Note that f ( x ) satisfies all the
requirements of Taylor's theorem. Therefore, we get
=0
If Q is an initial approximation to the root, then the corresponding point on the graph is
P (Q , f (xo)). We draw a tangent to the curve at P. Let it intersect the x-axis at T
(Figure 9.4). Let xl be the x-coordinate of T. Let S (a, 0) denote the point on the x-axis
where the curve cuts the x-axis. We know that a is a root of the equalion f (x) = 0. We
take xl as the new approximation which may be closer to a than no. Now let us find the
tangcnt at P (XQ, f (xo)). The slope of the tangent at P (xo , f (no)) is given by f' (no).
Therefore by the point-slope form of the expression for a tangent to a curve, we can write
This tangent passes through the point T (XI, 0) (Figure 9.4). Therefore we get
x ~ f (xo)
'
= xof' (xo) -f !XO)
i.c.,
This is the first iterated value. To get the second iterated value we again consider a
tangent at the point P (.XI, f (XI))on the curve (see Figure 9.4) and repeat thc process.
Then we get a point TI (x2 , 0) on the x-axis. From the figure, we obscrve that TI is more
close to S ( a , 0) than T. Therefore after each iteration the approximation is coming
closer and closer to the actual root. In practice we do not know the actual root of a givcn
function.
Let us take up some examples.
Example 9.3
Therefore xl = 1 -
0:4425922
- 1.425519
Foa i = 2, we get
X,
= 1.31048 -
2 - tan (1.3104s)
1 - tan2 (1.31048)
Sr'"ti"qof Mwbraieand
TramicedentalEquatiom
Sdution of Equatiom
Now x5 and
are correct to five decimal places. Hence we stop the iteration process
Example 9.4
Find an approximate value of c u i n g the Newton-Raphson formula.
Solution
A?
A?
Then f (x) satisfies all the conditions for applying Newton-Raphson method. We
choose 1-0 = 1 as the initial approximation to the root. This is because we know
that *lies between f i a n d f i a n d therefore we can assume that the root will be
close to 1.
The iteration formula is
Similarly
Thus the value of @correct to seven decimal places is 1.4142136. Now you can
check the value with the calculator.
Note 1 :The method used in the above example is applicable for finding square root of
any positive real number. For example, suppose we want to find an approximate value of
fiwhere A is a positive real number. Then we consider the equation - A = 0. The
iterated formula in this case is
A?
Supposef ' (xi) is zero in a neighbourhood of the root, then it may happen that
f ' (x,) = 0 for some x,. In thts case we cannot apply Newton-Raphson formula,
since division by zero is not allowed.
Remark 2
Another difficulty is that it may happen that f ' (x) is zero only at the roots. This
happens in either of the situations.
(i) f (x) has multiple root at a . Recall that a polynomial functionf (x) has a
multiple root a of order N if we can writs:
(a)
(b)
2
Then find a1 approximare mot of 2x .- 2 sin x = 0 in the interval [O, n] with
error less than 10- start wi t11 xg = 13.
(c) Using Newton-Raphson square root algorithm, find the following roots within
3
an accuracy of 10- .
(i)
= 3
Carl Newton-Raphson iteration method he used to solve the equation xlB = O'!
This clitelion gives US an idea of how many successive iterations have to be carried out to
obtain the root to the desired accuracy.
solution oPAlgebral~and
Transcendental Equations
Dtfinltlon 1
Let a , xl
.. . . ,xn
Xn
=0
. We say that
for some number h > 0.p is called the order of convergence and h is called the
asymptotic error constant.
xi
If this condition is satisfied for an iteration process then we say that the iteration
process converges linearly.
Setting n = 0 in the inequality (9.8), we get
Forn = 1,weget
(XnJ =
(9.9)
converges to themot.
Now we shall find the order of convergence of the iteration methods which you have
studied so far. Let us consider bisection method.
Convergence of Blsectlon Method
Suppose that we apply the bisection me& on the interval [UCJ, bol for the equation
f (x) = 0. In this method you have seen that we construct intervals
[ao, bol 2 [a1 , bll I> [a2 , b2] 2 . . . each of which contains the required root of the
given equation.
1 i.e.,
Recall that in each step the interval width is reduced by 2
and
We know that the equationf (x) = 0 has a root in [q,, bo]. Let a be the root of the
equation. Then a lies in al\ the intervals [ai, bi], i = 0.1.2, ,. . . . For any n, let
an + bn
c, = -denote the middle point of the interval [a,, b,J. Then co ,cl ,c2, . . . are
2
taken as successive approximations to the root a. Let's check the inequality (9.8) for
Thus
FnJ=
converges to the root a.Hence we can say that the bisection method
always converges.
For practical purposes, we should be able to decide at what stage we can stop the iteration
to have an acceptably good approximate value of a.The number of iterafions required to
achieve a given accuracy for the bisection method can be obtained. Suppose that we want
an approximate solution within an error bound of lo-" (recall that you have studied error
bounds in Unit 8). Taking logrithms on both sides of Equation (9.10), we find that the
number of iterations required, say n, is approximately given by
n = int
"I
where the symbol 'int' stands for the integral part of the number in bracket and I@, bo[
is the initial interval in which a root lies.
Let us work out an example.
Example 9.5
Suppose that the bisection method is used to find a zero off (x) in the interval
[O, 11. How many times this interval be bisected to guarantee that we have an
approximate root with absolute error less than or equal to 10Solution
Let n denote the requirednumber. To calculate n, we apply the formula (9.11) with
%=0,bo=1andM=5.
'.
Then
Solution oC Eqnatlons
--=int [16.60964047] = 17
This table shows that for getting an approximate value with an absolute error bounded by
10- we have to perform 17 iterations. Thus even though the bisection method is simple
to use, it requires a large number of iterations to obtain a reasonably good approximate
root. This is one of the disadvantages of the bisection method.
Note : The formula given in Equation (9.11)shows that, given an acceptable error, the
number of iterations depends upon the initial interval and thereby depends upon the initial
approximation of the root and not directly on the values off (x) at these approximations.
Next we shall obtain the convergence criteria for the secant method.
',
Let f (x) = 0 be the given equation. Let a denote a simple root of the equation f (x) = 0.
Then we have f' (a) # 0. The iteration scheme for the secant method is
&i
+ a. Substituting in Equation
(9.12),we get
Now we expand f (ei + a) and f (ei- a)using Taylor's theorem about the point x = a
We get
since f (a) = 0.
Similarly,
Therefore
f" (a)
)2ft(a)
Substituting Equation (9.14) and Equation (9116)in Equation (9.13),we get=f'(a)
&i-Ei-1
1[
1+
(E i - E i - 1
"]
(9.16)
2f
Ei
(a)
f ' (a)
"
"
(a)
.f " (a)
f ' (a)
B y neglecting the terms involving ~iE?-
+ &i2
This relationship between the errors is called the error equation. Note that this
relationship holds only if a is a simple root. Now using Equation (9.17) we will find
numbers p and h such that
Setting i = j - 1, we obtain
or
~i =
h E$'-
&;/'
hl/p Ei - 1
~.e.,
ei - = A- I4 EiVP
Combining Equations (9.17) and (9.18),we get
h&f =
EiEl-l
f" (a)
2 f ' (a)
Substituting the expression for ~i- 1 from Equation (9.19) in the above expression we get
~ion
Now, to get the number h, we equate the constant terms on both sides of Equation (9.20).
Then we get
Solution of Quatiom
r2 (air
Hence the order of convergence of the secant method is p = 1.62 and the asymptotic
error constant is
f" (a)
+'
Example 9.6
The following are the five successive iterations obtained by secant method to find
the root a = - 2 of the equation x 3 - 3x + 2 = 0.
= - 2.6, ~2 = - 2.4, ~3 = - 2.106598985,
XI
~4 =
32 ~ q ,
Solution
Letf(x) = 2 - 3 x + 2
Then
f'(x) = 3 2 - 3 , f P ( - 2 ) = 9
f" (x) =
[- 51
[- t]
a,
f"
(- 2) =
- 12
.618
merefore 1 =
.618
- 0.778351205
Now
ES
I x5 - a I
1 - 2.000022537 + 2 1
= 0.000022537
-and
~q
= 1 - 2.022641412 + 2 1 = 0.022641412.
= e5
Let a denote a root of the equation. Let xo be an initial approximation to the root. The
iteration formula is
x,+l = g ( x i ) , i = 0, 1 . 2 , .
..
(9.21)
We assume that g '(x) exists and is continuous and I g ' (x)1 < 1 in an interval containing
the root a. We also assume that xo , xl , . . . lie in this interval.
Since g '(x) is continuous near the root and I g ' (x)1 < 1, there exists an interval
l a - h , a + h [ , where h > 0, such that 1 g ' (x)I Ik for some k, where 0 < k < 1.
Since a is a root of the equation, we have
62
a = g (a).
(9.22)
.
Now the function g (x) is continuous in the interval 1x1 , a[ and g ' (x) exists in this
interval. Hence g (x) satisfies all the conditions of the mean value theorem (see Unit 8).
Then by the mean value theorem there exists a 6 between xi and a such that
Note that.5 lies in ]a- h , a + h[ and therefore 1 g ' (5) 1 < k and hence
-
This shows that the sequence of approximations (xi) converges tg a m v i d e d that the
initial approximation is close to the root.
We summarise the result obtained for this iteration process in the following Theorem.
Theorem 1
If g (x) and g ' (x) are continuous in an interval about a root a of the equation
x = g (x), and if I g ' (x) 1 < 1 for all x in the interval, then the successive
approximations xl , q , . . . given by
conqrges to the root a provided that the initial approximation is chosen in the
above interval.
all now discuss the order of convergence of this method. From the previous
we have the result.
Note that 5 is dependent on each xi . Now we wish to determine the constants X andp
independent of xi such that
Note that as the approximations xi get closer to the root a, g ' (5) approaches a constant
value g ' ( a ) . Therefore, in the limiting case, as i + -, the approximations satisfy the
relation
Solution of Equations
Hence, if g ' (a)= 0 and g " (a)f 0, then this iteration method is of order 2.
Example 9.7
al
(ax;
+ b)
> I pl.
XI
Solution
b
converge to x = a,then we require 1 g ' (a)1 = I - 7I < 1.
a
Thus
i.e..
Now you recall from your elementary algebra course that if a and P are the roots,
then
Therefore 1 bl = 1 a 1
-bl =
a1
l pl
Hence
I a1 > I P I
Finally we shall discuss the convergence of the Newton-Raphson method.
Convergence of Newton-Raphson Method
To obtain the order of the method we proceed as in the secant method. We assume that a
is a simple root off (x)= 0. Let
Then we have
i.e.,
Ei+l =
Ei f'
(Ei + a ) -f (Ei + a )
f'
( ~ +
i a)
Now we expand f ( E+~a ) and f' (ci + a ) , using Taylor's theorem, about the point a . We
have
0. Therefore
This shows that the errors satisfy Equation (9.6) with p = 2 and h =
-f"O ~ e n c e
2f' (a)'
Newton-Raphson method is of order 2 can be used. That is at each step, the error is
proportional to the square of the previous error.
An alternate mefhod to show that the order is 2 can be used. Note that we can write (9.24)
in the form x = g ( x ) where
Then
A[,
g'(x) = - x - %]=I-
Now,
gJ(a)=
*f"
(a) =
o ,sincef ( a ) = o andf'
( a ) t 0.
[f' (00l2
He ce by the conclusion drawn in the above Example, the method is of order 2. Note that
this is true only if a is a simple root. If a is a multiple root i.e., if f' ( a ) = 0, then the
convergence is not quadratic but only linear. We shall not prove this result, but we shall
illustrate this with an example.
Let us consider an example.
Example 9.8
Let f (x) = (x - 214 = 0. Starting with the initial approximation Q = 2.1, compute
the iterations xl , x2 , x3 and xq using Newton-Raphson method. Is the sequence
converging quadratically or linearly?
Solution
,= ;
(3xi - 2)
Starting with Q = 2.1, the iterations are given by
Similarly,
x2
2.05625
and
3
Thus the convergence is linear in this case. The error is reduced by a factor of 4
with each iteration. This result can also be obtained directly from Equation (9.25).
In rbe next section, we shall begin by listing some of the important results about the roots
of polynomial equations.
where a, a1 ,. . . ,a, are real numbers and a, # 0. You know that the roots of a
polynomial equation need not be real numbers, it can be complex numbers, that is
numbers of the form z = a + i b where a and b are real numbers. The following results
are basic to the study of roots of polynomial equations.
Theorem 2 (Fundamental Theorem of Algebra)
Let p (x) be a polynomial of degree n 2 1 given by Equation (9.26). Tben
p (x) = 0 has at least one root; that is there exists a number a E C such that
p (a) = 0. In fact p (x) ha,^ n complex roots which may not be distinct.
Theorem 3
Let p (x) be a polynomial of degree n and a is a real number. Then
p(x) = ( x - a ) q ( x ) = r
(9.27)
for some polynomial q (x) of degree n - 1 and some constant number r, q (x) and r
are called the quotient polynomial and the remainder respectively.
In particular, if a is a root of the equation p (x) = 0, then r = 0; that is (x - a )
divides p (x).
Then we get
How do we determine q (x) and r? We can find them by the method of synthetic
division of a polynomial p (x). Let us now discuss the synthetic division procedure.
Consider the polynomial p (x) as given in Equation (9.26).
- 1 and r is a constant.
Now, to find bo , bl , . . . ,b, we simplify the right hand side of Equation (9.29)
and compare the coefficients of 2 , i = 0.1,. . . ,n, on both sides. Note that
p (a) = bo. Comparing the coefficients we get
Coefficient of f
: a, = b, ,
coefficientoff-'
: a,-1 = b,-1-ab,,
Coefficientof x0
: q = bo - a ,
bn = a,
b,-1
= a,+l a b ,
bO = a O + a b l
Sdution of Equatiom
We count the changes in the sign of the coefficients. Going from left to right there
are changes between 1 and - 15, between - 15 and 7 and between 7 and - 11. q e
total number of changes is 3 and hence it can have at most 3 positive roots. Now
we consider
Here there is only one change between 1 and - 15 and hence the equation cannot
have more than one negative root.
We now give another theorem which helps us in locating the real roots.
Theorem 6
Let p (x) = 0 be a polynomial equation of degree n 2 1. Let a and b be two real
numbers with a < b. Suppose further that p (a) # 0 and y (b) # 0. Then,
(i)
if p (a) and p (b) have opposite signs, the equationp (x) = 0 has an odd number
(ii) if p (a) and p (b) have like signs, then p (x) = 0 either has no root or an even
number of roots between a and b.
Note :In this theorem multiplicity of the root is taken into consideration i.e., if a is a root
of multiplicity k it has to be counted k times.
8
Then
SAQ 3
(a) Consider the equation 2x - loglo x = 7 lies in ]3.71(.1.79[. Find the number n
of bisections required to have an approxjmate nlot witl! absolute error less than
or equal to 10- 7.
b
(b) For the equation given in Example 9.7, show that the iteration xi+ = -X; + a
will converge to the root x = a; when 1 o.1 < I I .
32 -c 3..
--.
--
1 ir: divided by x - 2 .
tic
check whcthcr fro - .? 1s n ro:tt ol the polynomial
( c ) I J : . : I s~~~ ~ ~ t h ccl~vjsion
t
1?x2
13 ==
0 *ilr~.lirrlrl thi:
7
(]td0t7"11
r x
in,ln, rregative rootc doc', the i3qu;~t~cr~l
i~
Also t l ~ ~ r n n ~tlw
n e~llirnh~r
of pi~sitiv-foots, it any
p014'1101~11;11.
+ 41-' + 1
J:oir,l
(P)
Sllow that the biquadratic ryuallon I? (a) = t -t I ' - 2x2 i- 4-x -- 24 = 0 has at
1easl.two real roots one posltive and the olller negative.
In the next section we shall discuss one of the'simple methods for solving polynomial
equations.
We shall now discuss the Birge-Vieta method for finding the real roots of a polynomial
equation. This method is based on an orlginal method due to two English mathematicians
Birge and Vieta. This method is a modified form of Newton-Raphson method.
Let % be an initial approximation to the root a. The Newton-Raphson iterated formula for
improving this approximation is
To apply this formula we should be able to evaluate both p,, (x) andq',, (x) at any xi .The
most natural way is to evaluate
pF(xi) = nu,
$-I
+ (n-
1) a,,-] # - 2
+ . .. +
2a2 q
+ a1
However, this is the most inefficient way of evaluating a polynomial, because of the
amount of computations involved and also due to the possible growth @f
round off errors.
Thus there is a need to look for some efficient method for evaluating pn (x) and p ', (x).
Let us consider the evaluation of p,, (x) and p ,' (x) at xo using Homer's method as
discussed in the previous section.
We have
where
and
bo = Pn (%) = ro
We have already discussed in the previous section how to find hi , i
(9.33)
1,2, . . . ,n.
Next we shall find the'derivativep ,' (Q) using Homer's method. We divide 9,(x - q~)
using Homer's method. That is, we write
(x) by
= 4,- l(%)
Hence the Newton-Raphson method (Equation 9.31) using (9.33) and (9.38), simplifies to
So~ution
Here the coefficients are a0 = 4 0 a1 = 4 , a2 = 8 , a3 = 8 , a4 = - 6 and
as = 1. To compute b ~we
, form the following table.
1
/
Table 9.8
2 = ~(2)'
x4 = ~
( 2etc.
)
Thus each term c p takes two multiplications for k > 1. Then the total number of
multiplications involved in the evaluation of p (x) is 1 + 2 + 2 + 2 + 2 = 9.
When we use Homer's method the total number of multiplications is 5. The
number of additions in both cases are the same. This shows that less computation
is involved while using Homer's method and thereby reduces the error in
computation.
Let us now solve some problems using Birge-Vieta method.
Example 9.11
Use Birge-Vieta method to find all the positive real roots, rounded off to three
decimal places, of the equation
Now we evaluate p4 (0.5) and p( (0.5) using Homer's method. The results are
given in the following table.
Table 9.9
7
24
- 15
Therefore xz = 0.7459
24
- -----2.3132 - 0.6998
50.1469
Table 9.11
Table 9.12
- 15
Solution of Algebraic a d
TramwendentalEquntions
Since x3 and x4 are Lhe same, we get I x4 - x3 1 < 0.0001 and therefore we stop the
iteration here. Hence the approximate value of the root rounded off to three
decimal places is 0.698.
Next we shall illustrate how Birge-Vieta's method helps us to find all real roots of
a polynomial equation
Consider Equation (9.29)
+bn-1
Y - ~ + . .+.b l
where b 'i s are obtained by using xn and the Homer's method. This polynomial is
called the deflated polynomial or reduced polynomial. The next root is now
obtained using q, - 1 (x) and not p, (x). Continuing this process, we can
successively reduce the degree of the polynomial and find one real root at a time.
Let us consider an example.
Example 9.12
Find all the roots of the polynomial equation y3 ( x ) = + x - 3 = 0 rounded off
to three decimal places. Stop the iteration whenever 1 xi + 1 -xi I < 0.0001.
Solution
The equation93 (x) = 0 has three root. Since there is only one change in the sign.
of the coefficients, by Descarte's rule of signs the equation can have at most one
positive real root. The equation has no negative real root since p3 (- x) = 0,has no
change of sign of. coefficients.
Since p3 (x) = 0 is of odd degree it has at least one
,
real root. Hence the given equation x3 + x - 3 = 0 has one positive real root and a
complex pair. Since p (1) = - 1 and p (2) = 7, by intermediate.value theorem the
equation has a real root lying in the interval ] 1 , 2 [ .Let us find the real root using
Birge-Vieta Method. Let the initial approximation be 1.1.
Rrst Iteration
Table 9.13
-0'569 - 1.22289
Therefore xl = 1.1 - 4.63
Similarly, we obtain
X,
= 1.21347
Since 1 x2 - x3 1 < 00.001, we stop the iteration here. Hence the required value of
the root is 1.213, rounded off to three deciinal places. Next let us obtain the
deflated polynomial of ps (x). To get the deflated polynomial, we have to find the
polynomial q 2 (x) by using the final approximation^^ = 1.213 (see Table 9.14).
Table 9.14
Note thatp3 (1.213) = - 0.0022. That is, the magnitude of the error in satisfying
p3 (x3) = 0 is 0.0022.
Hence the three roots of the equation rounded off to three decimal places are
1.213.0.6065 + 1.4505 i and 0.6065 - 1.4505 i.
Remark
We now know that we can determine all the real roots of a polynomial equation
using deflated polynomials. This procedure reduces the amount of computations
also. But this method has certain limitations. The computations using deflated
polynomials can cause unexpected errors. If the roots are determined only
approximately, the coefficients of the deflated polynomials will contain some
errors due to rounding off. Therefore we can expect loss ot accuracy in the
remaining roots. There are some ways of minimising this error. We shall not bc
going into the details of these refinements.
SAQ 4
(a) Using syi~theticdivision, show that 2 is a simple root of the equation
(c)
Find the real root rounded off to two decimal places of the equation
x4 - 4 2 - 3x + 22 = 0 lying in the interval ] 2 , 3 [ by Birge-Vieta method.
solution of Eqaotiom
9.7
HORNER'S METHOD
This is a very elegant method of finding the real root of a numerical polynomial equation.
The method consists in first determining the integral part of the root and then computing
the decimal digits successively upto a desired degree of accuracy. Suppose it is required
to find an approximate real positive root of the polynomial equationf (x) = 0. Choose
two successive integers a and a + 1 in such a way that f (a) andf (a + 1) are of opposite
signs. Then a root of the equationf (x) = 0 lies between a and a + 1 and 'a' is taken as
the integral part of the root.
Let the root be now assumed as a . b e d e . . . where b e d e . . . are the digits in the
decimal part of the root. To build up the decimal part we proceed as follows :
(i) Diminish the roots of the given equation by 'a' so that the root of the
transformed equation corresponding to a is o . b e b e . . . .
(ii) Then multiply the roots of the new equation by 10 and consequently the roots of
this equation will be b . c d e . . . . By trail find two consecutive integers b and
b + 1 such that the root of this new equation lies between these two values. This
determines the digit b.
(iii) Diminish the root of the equation obtained in (ii) above by b and multiply the
root of the resulting equation by 10 and determine c, exactly in a similar fashion
as done earlier. The iteration process is continued till the value of the root upto a
desired accuracy is obtained.
We shall now illustrate this method with an example.
Example 9.13
Use Homer's method is find the root correct upto three decimal places of the
equation 2 - 3x + 1 = 0, theT00t lies between 1 and 2.
Solution
By Descarte's rule of signs, there are two positive roots and one negative root of
the given equation.
Also f (0) = 0 - 0 + 1 = 1, apositive
f(1) = 1 - 3 + 1 = - 1, anegative
f(2) = 8 - 6 + 1 = 3, apositive
Thus there are two positives roots.lying between (0, 1) and (1, 2). But we have to
determine the root between 1 and 2 only.
The integral part of the root is 1. To build up the decimal part, diminish the root of
the given equation by 1
1.5320
Root
It will be clear by trial that the root of this equation fi(x)= 0 will lie between 5
and 6 as fl(5)is -ve and fl (6)is +ve.
h e first digit of the decimal part is 5
Diminish roots of fl(x) = 0 by 5
Again by trial we notice that the root of f2 (x) = 0 lies between 3 and 4 as f2 (3)is
-ve and f2 (4)is +ve.
The digit at the second decimal position is 3.
Diminish the roots of f2 (x) = 0 by 3.
Again by trial we find that a root off3 (x) = 0 lies between 2 and 3 as f3 (2)is -ve
and f3 (3)is + ve.
The digit at the third decimal position is 2
Diminish the roots of f3 (x) = 0 be 2.
sdubonofAlgebraicand
Transcedentd Equations
Solutiou of Equations
Again by trial we find that the root of f4 (x)= 0 lies between 0 and 1 as f4 (0)is
-ve and f4 (1)is +ve. This gives the digit at the fourth decimal position as 0.
Thus the value of the root lying between 1 and 2 correct upto three decimal places
is 1.532.
Remark
We have observed that in this transformed equations f i (x)= 0,f2 (x)= 0,
f3 (.x) = 0,andf4 (x)= 0.We searched for two consecutive integers, using the
method of hit and trial, between which the root of the equation is to lie. The
computational work involved in determining these numbers is enormous. However,
it is worth noting that after a couple of transfdrmations the root of the transformed
equation can be conveniently located by considering its two lowest degree terms.
Thus in the transformed equation f2 (x)= 0,the coefficients of the last two terms
are dominant and therefore, omitting the first two terms and solving the residual
equation, we obtain
This gives
Similarly in f3 (x)= 0,the two lowest degree terms when equated to zero give,
4022708c - 8423000 = 0
= 4022700
:.
This is a method in which the curve y = f (x)is approximated by a second degree curve
very near to the actual root. The roots of the quadratic are considered as close
approximations to the root of the equationf (x)= 0.The method is iterative and
converges quadratically. The general iterative formula of the method is derived as follows.
Let xi- 2 ,.xi - 1 and xi be the three initial ap@roximationsto a root off (x)= 0,and let the
corresponding values of y = f (x)be denoted as y; - 2, yi, 1 and Vi. We asshme that a
parabola
y = a x2 + b x + c
i ) .
This gives
+ bxi-2 + c
+ bxi- 1 + c
2 + bx; + c
yi-2 = a 2 - 2
2
yi- 1 '= axi1
y; = a
I Y
(x-xi)
(xi- -.Xi)
Yi- 1
It is clear from the above equation that p which depends on x is an unknown quantity and
we need to determine it in order to get a better approximation of the root. Further, since x
and X i differ by a very small quantity, the magnitude of p is very small. Setting y = 0 in
the abvoe equation, we get a quadratic equation in p whose roots determine the value of
p. Thus we obtain
where
i
P
Sdotion of Equatiom
Hence
Equation (9.50) gives a better approximation of the root. It shouiibe remarked that the
sign in (9.50) should be chosen in a manner that the denominator becomes very large in
magnitude so that p converges to the root.
Example 9.14
Using Muller's method, find a root of the equation 2 - x - 1 = 0; starting with
xi - 2, xi - 1, $ as 0, 1 and 2 respectively.
Hence x = xi
+ p (xi- Xi -
= 2 - 0.3823594 = 1.6176406
.-.Error = 1
In the previous section we have discussed a method for finding real roots of polynomial
equations. In the next section we shall discuss a direct method for solving polynomial
equations. This method was developed independently by three mathematicians Dandelin,
Lobachevsky and Graeffe. But Graeffe's name is usually associated with this method.
The advantage of this method is that it finds all roots of a polynomial equation
simultaneously, the roots may be real and distinct, real and equal (multiple) or complex
roots.
where >> stands for 'much greater than'. Then we can obtain the roots approximately
from the coefficients of the polynomial equation as follows :
I p1I
an- 1
an
s d ~ t i ~m
oa
f ions
SOthe problem now is to find from the given polynomial equation, a polynomial equation
whose roots are widely separated. This can be done by the method which we shall
describe now.
In the present course we shall discuss the applications of the method to a polynomial
equation whose roots are real and distinct.
Let al, 012, . . . ,anbe the n real and distinct roots of the polynomial equation of degree n
given by
where ao, a l , a2, . . . ,an - an are real numbers and an # 0. We rewrite Equation (9.54)
by collecting all even terms on the one side and all odd terms on the other side, i.e.,
Now we expand both the right and left hand sides and simplify by collecting the
coefficients. We get
b o + b l y + b 2 y2
+ . . . + b,=O
(9.57)
where
bo =
&
bl = a12 - 2 % a 2
b2 = a$
- 2 a1 a3 + 2 a0 a4
ao
a1
a2
a3 . . .
a3
'a?
al
a3
-2aoa2
-2a1 a3
-2a2w
2aoa
- 2 a1 as
-2aoa
bo
bl
b2
b3 . . .
an
.
a8
To form Table 9.15 we first write the coefficients ao, a l , a2, . . . ,a, as the first row. Then
we form (n + 1) columns as follows.
The terms in each column alternate in sign starting with a positive sign. The first term in
each column is the square of the coefficients ak, k = 0,1,2, . . . ,n. The second term in
each column is twice the product of the nearest neighbouring coefficients, if there are any,
with negative sign; otherwise put it as zero. For example, the second term in the first
column is zero and second term in the second column is - 2 a0 a2. Likewise the second
term of the (k + 1)th column is - 2 ak- 1 ak + 1. The third term in the (k + 1)th column is
twice the product of the next neighbouring coefficients ak - 2 and ak+ 2, if there are any,
otherwise put it as zero. This procedure is continued until there are no coefficients
available to form the cross products. Then we add all the terms in each column. The sum
gives the coefficients bk for k = 0, 1, 2, . . . , n which are listed as the last term in each
column. Since the substitution x2 = - y is used, it is easy to see that if a l , 012, . . . , a, are
the n roots of Equation (9.54), then - a:, a;, . . . , a: are the roots of Equation (9.57).
Thus, starting with a given polynomial equation, we obtained another polynomial
equation whose roots are the squares of the roots of the original equation with neagive
sign.
We repeat the procedure for Equation (9.57) and obtain another.equation
whose roots are the squares of the roots of Equation (9.57) with a negative sign i.e., they
are fourth powers of the roots of the original equation with a negative sign. Let this
procedure be repeated n times. Then, we obtain an equation
72, . .
. ,b are given by
Now, since all the roots of Equation (9.54) are real and distinct, we have
I
Hence
lyl/
We conclude that if the roots of Equation (9.54) are distinct then for large m, the 2mth
powers of the roots are widely separated.
We stop this squaring process when the cross product terms become negligible in
comparison to square terms.
Since roots of Equation (9.58) are widely separated we calculate the absolute values of
the roots y l , 72, . . . ,y, using Equation (9.53). We have
The magnitude of the roots of the original equation are therefore given by
sdutiOnOfAlgebraic=d
Tramcendental Equations
This gives the magnitude of the roots. To determine the sign of the roots, we substitute
these approximations in the original equation and verify whether positive or negative
value satisfies it.
We shall now illustrate this method with an example.
Example 9.15
The equation has no negative real roots. Let us now apply the root squaring method
successively. Then we get the following results :
First Squaring
Table 9.16
Applying the squaring method to the new equation we get the following results.
\
Second Squaring
, Table 9.17
2 + 68332 +
1 7 8 8 6 8 8 ~+ 26873856 = 0
Table 9.18
a3 of
Since the equation has no negative real roots, all the roots are positive. Hence the
roots can be taken as 9.0017.4.0011 and 1.9990. If the approximations are rounded
to 2 decimal places, we have the roots as 9.4 and 2. Alternately, we can substitute
the approximate roots in the given equation and find their sign.
sday5
'
i i i
iiL.
;pth of x :ileiers given by tile t..qrl;lti;jn
I
"I-' -
3.12'
+ 2.5
.= i)
Sdotion of Equations
ic. )
I?i~!.i~n~rinc
all rct :!s i:,f tlre A:!' ion;ing equatio~tsi,y i..ir:~r(tc''sroot squarii.rg
x~1ell1c~9.
u.;rliC fllrr'-esq~~ariags.
(f)
($9
r ~nci
.-
6 J' t 1 l x - 6 = 0, using
7
- 5 .r. -
?- - O. wing Graeftc's
9.10 SUMMARY
.
- xif (xi- 1)
, i = 0,1,2 , . . .
1)
where q and x are any two approximations of the root.
(iii) Newton-Raphson Method
The iteration formula is
Xi+l =
xi-
l f (xi)
f (xi)
- f (xi-
We have to fust fmd two numbers a and h such that f (a) f (h) < 0. Since
the function loglo x is defined only for positive values of x, we consider
only positive numbers x. Let us take x = 1 , 2 , 3 , . . . .
Then,
This shows that f (2) f (3) < 0 and therefore a root lies in ] 2 , 3 [ .Now put
a = 2 and h = 3. Then the fust approximation of the root is
= 2.7406
Since x2 and x3 rounded off to three decimal places are the same, we stop
the process here. Hence the desired approximate value of the root rounded
off to three decimal places is 2.740.
(ii) Let f (x) = x sin x - 1
Sincef (0) = - 1 and f (2) = 0.818594854, a root lies in the interval
10, 2[.The first approximation is
andf ( x l ) =
- 0.02001921
Sincef (xl) < 0 andf (2) > 0, the root lies in 11. ~ 7 5 0 1 7 , 2 1 .
mesecond approximationis
x2
= 1.212407 4
as
and
sOlutiondEqaatioos
f (x2) =
- 0.00983461
= 1.11415714
Since xg and x4 rounded off to three declrnal places are the same, we stop
the process here. Hence the desired root is 1.114.
(b) L e t f ( x ) = 2 x - l o g l o x - 7 = 0
f (3.5) =
The exact root lies between 3.7888 and 3.79. The second approximationis
given by
(c)
Important Note :In the calculation of the second approximation above, the
interval has been narrowed down by taking the end values as 3.7888 and 3.79
instead of the values 3.7888 and 4. This has been suggested by the fact that
functionf (x) changes sign when x is changed from 3.7888 to 3.79 and
therefore, a root off (x) = 0 lies between 3.7888 and 3.79. This procedure can
be used in successive steps as this saves both time and labour and conveyance
to exact root becomes faster.
x = 0.6822.
and
f ( ~ 2 )= 0.0145682
= 2.41421384 and
xs = 2.41421356
Since x4 and x5 rounded off to 5 decimal places are the same, we stop the
process here. Therefore the required root rounded off to 5 decimal places is
2.41421.
Using a calculator we
Now we compare this root with the exact root 1 + 6.
find 1 + fi = 2.41421, rounded off to five decimal places. Hence the
computed root and exact root are the same when we round off to five decimal
places.
(i) We first note thatf (1) 4 0 and f (2) > 0. Therefore a root lies in the open
interval]1,2[. The first approximationxl is
Interval
Approximation X i
11, 21
1.57142
(xi)
- 1.36449
The table shows that x5 and x4 are correct to three decimal places.
Therefore we stop the process here. Hence the root correct to three decimal
places is 1.731.
(ii) In secant method we start with two approximations a = 1 and b = 1.
Then the first approximation is the same as in part (i), namely
To calculate the next approximation x;?we take b and XI. Here also we are
getting the same value as in part (i),namely
x;? = 1.70540
Then we take XI = 1.57142 and x;? = 1.70540 to get the third
approximation x3. We have
and
xs = .73205
Since x4 and q rounded off to three decimal places are the same, we stop
here. Hence the root is 1.732, rounded off to three decimal places.
Let us now compare the two methods. We fist note that 1 y + I - xi 1 gives
the error after ith iteration
In regula-falsi method, the error after 5th iteration is
= .ooO54
whereas in secant method, the error after 5th iteration is
I xS-x4(
1.73205-1.731991
This shows that the error in the case of secant method is smaller than that
in regula-falsi method for the same number of iterations.
The given functionf (x) =
continuous everywhere.
- 4 are
and
f ( X I ) = ( 0 . 2 5 ) ~- 4 (0.25)
1 = 0.015625
= 0.254098
Similarly, we get
Since x2 and x3 rounded off to four decimal places are the same, we stop the
iteration here. Hence the root is 0.2541.
(b) Letf (x) = 2x - 2 - sin x. Thef (x) and f' (x) are continuous everywhere.
Starting with xg = 1.5, we compute the iterated values by the Newton-Raphson
formula. The fust iteration is
= 1.5 -
1 - sin (1.5)
2 - cos (1.5)
Similarly,
q = 1.498701
We find 1
x2
Since I x3 - x2 I <
root is 2.8284.
Since I
x3
- x2 I
<
(d)
(x)
312"
SAQ 3
(b) Here g ( x ) =
. The iteration
-x+a
particular we require
+ a)2
I
i.e.,
But we have a
gP(a)l = I
-I <
( a+ a12
(a+a12 c I b I .
f3 = - a and a f3 = b. Therefore we get
p2 > I bl = I al I pl
I al < I P I
i.e.,
(c) The iterated formula is
1.458333333
2x3 - 5x2 + 3 x -
Hereu3 = 2, a;! = - 5, al = 3,
this case i$ as follows :
2x2 - x +
(e) From the Homer's table in this case. From the table you can see that the last
term in the 3rd row is zero. Hence 3 is a root of the equation. The quotiept
polynomial is? + 4x2 - x - 4.
(f)
(g)
SAQ 4
+ 3x - 4
= 0.
-2
Thereforex1 = - 2 - lo - 1.796.
- 49
Repeating the procedure to find x2, we have
- - 32.565
= 1.7425
i0.4
Since there is only one change in the sign of the coefficients of p (x),the
equation has at most one real root. The equation has no negative teal root since
there is no change in the sign of the coefficients of p (- x). Also
and
- 1.b73
- 1.13359 i.
( f ) 2.05.
SAQ S
(a) L e t f ( x ) =
x? -
3x2
+ 2.5
f(1) = 1 - 3
f(2) = 8
+ 2.5
12
= 0.5 > 0
+ 2.5
= -1.5 < 0
Since the last two terms are dominant, omitting the first term, the approximate
value of x is
.I? +
-> 6 and
297
< 7,
~ ofAlpebdcand
a
~
~
Tr~nsceodedalEquPtMM
(ii) 2.73
(iii) 3.18
(c) To start with we choose
xi-2 = 1 , xi-1 = 2 and xi = 3
The corresponding values of y are
Xi-2
- 1 . X,-1
= 0. Xi = 1
1
Solving for - which is given by
1
We get two values of -. For the numerator to be largest in magnitude. We take
- ve sign and the value of p after substituting for the quantities is given by
Therefore the next approximation to the root is given by
(ii) Let y =
x?
2-x
6 2 - 36x
+ 40
= 0
First squaring
Second squaring
Third squaring
2560000
320256
10032
.65536 X 1013
-10256 X 1012
.lo064 X lo9
.513638 x 10"
- .640512 x lo6
1o8
.65536 x l0l3
2+
yli = lo8
.65536 x 1013 = 0
~ubstitutin~
the computed values in the original equation, we get that the roots
are approximately - 10,2.18 and 1.83. Therefore the roots are - 10,2 and 2.
(ii) Computed values of the roots are 3.014443,1991424 and 0.9994937.
(iii) Computed values of the roots are 7.017507, - 2.974432,0,9581706.
(f)
+ 5x
2=0
First squaring
33
18
Second squaring
16
Third squaring
Then