4.

2 Regula-falsi method or method of false position

This method is the oldest method for finding the real root of an equation f (x) = 0 . It is also known

as method of chords or method of linear interpolation. Like the bisection method, the false

position method starts with two points a0 and b0 such that f (a0 ) and f (b0 ) are of opposite signs,

which implies by the intermediate value theorem that the function f has a root in the interval

[a0, b0], assuming continuity of the function f.

We first determine by the method of tabulation a sufficiently small interval [a0 , b0 ] containing the

only root α of the equation f (x ) = 0 such that f (a0 )f (b0 )< 0 .

Then, we approximate the portion of the curve y = f (x ) between the two points (a0 , f (a0 )) and

(b0 , f (b0 )) by a straight line. It has been shown in fig. 3.

Fig 3 graphical representation of regula-falsi method

Now the equation of the chord joining the two points (a0 , f (a0 )) and (b0 , f (b0 )) is given by

f (b0 )− f (a0 )(
y − f (b0 ) = x − b0 ) (4.2.1)
b0 − a0

1
The method consists in replacing the portion of the curve between the points (a0 , f (a0 )) and

(b0 , f (b0 ))by means of the chord joining these two points, and taking the point of intersection of
the chord with the x-axis as an approximation to the root. In this case, the point of intersection is

obtained by putting y = 0 in eq. (4.2.1). Thus, the first approximation x1 to the root α is obtained

as

x1 = a0 f (b0 )− b0 f (a0 ) (4.2.2)

Equivalently we may write,

a0 f (b0 ) + b0 f (a0 )
x1 = (4.2.3)
f (b0 ) + f (a0 )

Next, we compute f (x1 ).

If f (a0 ) and f (x1 ) are of opposite signs i.e. f (a0 )f (x1 ) < 0 , then the root lies in [a0 , x1 ] ,

so we set a1 = a 0 and b1 = x1 .

Otherwise, if f (x1 ) and f (b0 ) are of opposite signs i.e. f (x1 )f (b0 )< 0 , then the root lies in [x1, b0 ] ,

we set a1 = x1 and b1 = b0 , so that in either case [a1, b1] contains α , i.e. f (a1) f (b1) < 0 .

We repeat the above step to obtain second approximation x 2 , third approximation x3 and so on.

In general, having obtained the interval [an ,bn ] containing α such that f (an ) f (bn ) < 0 . We obtain

the (n+1)-th approximation xn +1 as

2
 bn − an
xn+1 = an − f (an )
f (bn ) − f (a n )

b f (a ) + a f (b )
n n n n
=
f (an ) + f (bn )

Now, we set an+1 = an and bn+1 = xn+1 if f (an ) f (xn+1) < 0 .

Otherwise, we set an+1 = xn+1 and bn+1 = bn if f (xn+1) f (bn ) < 0 , so that [an+1, bn+1] contains α . i.e.

f (an+1 ) f (bn+1 ) < 0 .

Example 4:

Find the real root of the equation x log10 x −1.2 = 0 correct to four decimal places using regula-falsi

method.

Solution:

We first apply method of tabulation in order to find the location of rough value of the root.

Table 3: location of the root

x f (x )

1 -1.2

2 -0.6

3 0.23

We note that f(2)<0 and f(3)>0. Thus the given equation changes its sign within the interval [2,

3]. Therefore, there exists at least one real root of the equation within [2, 3].

3
 b f (a ) + a f (b )
n n n n
Now, for the regula-falsi method, we compute x n+1 = .
f (a n ) + f (bn )

If f (an ) f (xn+1 ) < 0 , then the root lies between [an , xn+1 ] i.e. we set an+1 = an and bn+1 = xn+1 .

Otherwise, if f (xn+1 ) f (bn ) < 0 , then the root lies between [xn+1 ,bn ] i.e. we set an+1 = xn+1 and

bn +1 = bn . The successive iterations have been presented in Table 4.

Table 4: Table for finding real root

n an bn f (an ) f (bn ) bn f (an ) + an f (bn ) f (x n+1 )

0 2 3 -0.59794 0.231364 2.72101 -0.0170911

1 2.72101 3 -0.0170911 0.231364 2.74021 -0.000384056

2 2.74021 3 -0.000384056 0.231364 2.74064 -8.58134E-6

3 2.74064 3 -8.58134E-6 0.231364 2.74065 -1.91717E-7

Hence, the required real root is 2.7406 correct to four decimal places.

Example 5:

Find the real root of the equation x e x − cos x = 0 correct to four significant figures using regula-

falsi method.

Solution:

We first apply method of tabulation in order to find the location of rough value of the root.

4
 Table 5: location of the root

x f (x )

0 -1

0.5 -0.0532219

0.6 0.267936

1 2.17798

We note that f(0.5)<0 and f(0.6)>0. Thus the given equation changes its sign within the interval

[0.5, 0.6]. Therefore, there exists at least one real root of the equation within [0.5, 0.6].
bn f (an ) + an f (bn )
Now, for the regula-falsi method, we compute xn+1 = .
f (an ) + f (bn )

If f (an ) f (xn+1 ) < 0 , then the root lies between [an , xn+1 ] i.e. we set an+1 = an and bn+1 = xn+1 .

Otherwise, if f (xn+1 ) f (bn ) < 0 , then the root lies between [xn+1 ,bn ] i.e. we set an+1 = xn+1 and

bn +1 = bn . The successive iterations have been presented in Table 6.

Table 6: Table for finding real root

5
 n an bn f (an ) f (bn ) bn f (an ) + an f (bn ) f (x n+1 )
xn+1 =
f (a ) + f (b )
n n

0 0.5 0.6 -0.0532219 0.267936 0.516572 -0.00360274

1 0.516572 0.6 -0.00360274 0.267936 0.517679 -0.000238932

2 0.517679 0.6 -0.000238932 0.267936 0.517752 -0.0000158242

3 0.517752 0.6 -0.0000158242 0.267936 0.517757 -1.04793E-6

Hence, the required real root is 0.5178 correct to four significant figures.

4.2.1 Order of convergence of regula-falsi method

It may be shown that the error at the (n+1)-th step is related to the error in the n-th step by the

expression

ε n+1 ≅ A ,
εn (4.2.1.1)

where A is a constant depending on the function f . This shows that the sequence of successive

iteration {xn } converges linearly to the root.

4.2.2 Advantage and disadvantage of regula-falsi method

• Advantage

This method is very simple and does not require to calculate the derivative of f (x) . Moreover,

this method is certainly convergent. Since the solution remains bracketed at each step,

convergence is guaranteed as was the case for the bisection method.

• Disadvantage

6
The method is first order and is exact for linear f. The method is very slow since it converges

linearly. Also, the initial interval in which the root lies is to be chosen very small.

4.2.3 Algorithm for regula-falsi method

Step 1: Start the program.

Step 2: Define the function f (x) .

Step 3: Enter the interval [a, b] in which the root lies.

Step 4: If f (a) f (b) < 0 . Then go to step 5 else step 9.

a f (b) + b f (a)
Step 5: Calculate x = .
f (a) + f (b)

Step 6: If f (a) f (x) < 0 , set b = x , otherwise if f (x) f (b) < 0 set a = x .

Step 7: If a − b < ε , ε being the prescribed accuracy. Then go to step 8 else 5.

Step 8: Print the value of x which is required root.

Step 9: Stop the program. □

7
4.3 Fixed-point iteration

Let, [a0 , b0 ] be an initial small interval containing the only root α of the given equation

f ( x) = 0 .

Let us rewrite given equation as

x = φ (x) (4.3.1)

so that the root α satisfies the equation

α = φ (α ) (4.3.2)

We assume that, φ(x) is continuously differentiable sufficient number of times in [ a0 , b0 ] such

that for x ∈ [ a0 , b0 ] , φ ( x) ∈ [a0 , b0 ] .

We take the successive approximation using the formula

xn +1 = φ ( xn ) , n ≥ 0 , (4.3.3)

starting with x0 = a0 or b0 .

Substracting eq. (4.3.3) from eq. (4.3.2), we get

α − xn +1 = φ (α ) − φ ( xn ) = (α − xn )φ ′(ξ n ) , (by using Lagrange’s mean value theorem)

where min{α , xn } < ξ n < max{α , xn }

Therefore, ε n +1 = ε nφ ′(ξ n ) , where min{α , xn } < ξ n < max{α , xn } .

This is called error equation.

1
4.3.1 Condition of convergence for fixed-point iteration method

Theorem 3: Let α be the root of the equation f ( x) = 0 i.e. x = α be a solution of x = φ (x) and

suppose φ ( x ) has a continuous derivative in some interval [a 0 ,b0 ] containing the root α . If

φ ′( x ) ≤ K < 1 for all x ∈ [a 0 , b0 ] , then the fixed point iteration process xn +1 = φ (xn ) converges with

any initial approximation x 0 ∈ [a 0 , b0 ] .

Proof:

According to the hypothesis of the theorem

α = φ (α ) (4.3.1.1)

We know the iteration scheme of fixed-point method is

xn +1 = φ (xn ) (4.3.1.2)

By subtracting (4.3.1.2) from (4.3.1.1), we have

α − xn +1 = φ (α ) − φ (xn )

Since, ε n +1 = α − x n +1 is error at the (n+1)-th approximation.

So, ε n +1 = (α − xn )φ ′(ξ ) , applying Lagrange’s mean value theorem

where min{α , xn } < ξ < max{α , xn }

Therefore, ε n+1 = α − xn φ ′(ξ ) ≤ K α − xn ≤ K ε n (4.3.1.3)

Again it can be written as

2
 ε n ≤ K ε n−1

Therefore,

ε n ≤ K 2 ε n − 2 ≤ K 3 ε n − 3 ... ≤ K n ε 0

Now, since K < 1 , K n → 0 as n → ∞

So, ε n → 0 as n → ∞

This implies

α − x n → 0 as n → ∞

Therefore, x n → α as n → ∞ .

Hence, the condition of convergence of the fixed-point iteration is that K < 1 or φ ′( x) < 1 in

[a0 , b0 ] .

Corollary:

It can be observed that the condition of convergence of fixed point iteration method is φ ′( x) < 1

in the neighbourhood of α . This fact can be easily examined from the following fig. 4:

3
 −1 < φ ′( x) ≤ 0 ⇒ convergence −1 < φ ′( x) ≤ 0 ⇒ divergence

0 < φ ′( x) ≤ 1 ⇒ convergence 1 < φ ′( x) ⇒ divergence

Fig. 4 The conditional convergence of fixed point iteration method

Example 6:

Find the real root of the equation x 3 + x − 1 = 0 correct to three significant figures using fixed-

point iteration method.

Solution:

Let f ( x) = x 3 + x − 1 = 0 .

We first apply method of tabulation in order to find the location of rough value of the root.

Table 7: location of the root

0 -1

1 1

We note that f(0)<0 and f(1)>0. Thus the given equation changes its sign within the interval [0,

1]. Therefore, there exists at least one real root of the equation within [0, 1].

Now, we rewrite the equation x 3 + x − 1 = 0 as

1
x= 2
= φ ( x) , say
x +1

− 2x
Hence, φ ′( x) = so that φ ′( x) < 1 in 0 < x < 1 . Therefore, according to theorem 3, iteration
( x 2 + 1) 2

in fixed-point method certainly converges.

We choose initial approximation x0 = 0 .

The successive iterations generated by the eq. (4.3.3) have been presented in Table 8 and it has

been cited graphically in fig. 5.

5
 Fig. 5 successive iterates in fixed point iteration

Table 8: Table for finding real root

n xn xn +1 = φ ( xn ) f (x n +1 )

0 0 1 1

1 1 0.5 -0.375

2 0.5 0.8 0.312

3 0.8 0.60976 -0.163535

4 0.60976 0.72897 0.116337

5 0.72897 0.653 -0.0685556

6 0.653 0.70106 0.045624

7 0.70106 0.67047 -0.0281294

8 0.67047 0.68988 0.0182119

9 0.68988 0.67754 -0.011432

10 0.67754 0.68537 0.0073189

11 0.68537 0.68039 -0.00462747

12 0.68039 0.68356 0.00294911

13 0.68356 0.68155 -0.00187003

14 0.68155 0.68282 0.00118959

15 0.68282 0.68201 -0.000755204

Hence, the required real root of the given equation is 0.682 correct to three significant digits.

6
Example 7:

Find the real root of the equation 10 x + x − 4 = 0 correct to six decimal places using fixed-point

iteration method.

Solution:

Let f ( x ) = 10 x + x − 4 = 0 .

We first apply method of tabulation in order to find the location of rough value of the root.

Table 9: location of the root

x f (x)

0 -3

1 7

We note that f(0)<0 and f(1)>0. Thus the given equation changes its sign within the interval [0,

1]. Therefore, there exists at least one real root of the equation within [0, 1].

Now, we rewrite the equation 10 x + x − 4 = 0 as

x = log10 (4 − x) = φ ( x) , say

−1
Hence, φ ′( x) = so that φ ′( x) < 1 in 0 < x < 1 . Therefore, according to theorem 3, iteration
(4 − x ) ln 10

in fixed-point method certainly converges.

We choose initial approximation x0 = 0 .

7
The successive iterations generated by the eq. (4.3.3) have been presented in Table 10.

Table 10: Table for finding real root

n xn xn +1 = φ ( xn ) f (x n +1 )

0 0 0.60205999 0.60206

1 0.60205999 0.53121571 -0.0708443

2 0.53121571 0.54017729 0.00896159

3 0.54017729 0.53905384 -0.00112345

4 0.53905384 0.53919484 0.000140998

5 0.53919484 0.53917715 -0.0000176934

6 0.53917715 0.53917937 2.22033E-6

7 0.53917937 0.53917909 -2.78627E-7

Hence, the required real root of the given equation is 0.539179 correct to six decimal places.