CHAPTER 2 ROOT FINDING

2.0 INTRODUCTION
• The root-finding problem is one of the most basic problems of numerical approximation. In
many practical problems we are required to solve an equation of the form 𝑓 𝑥 = 0. If f (x) is
purely a polynomial in x then the equation f (x) = 0 is said to be an algebraic equation,
otherwise if f (x) also contains other functions like Exponential, Logarithmic, Trigonometric
functions in x , then the equation is called a Transcendental Equation.

• NB: In numerical analysis all the calculation are carried out in radians mode not in degrees
mode.

• A result which will prove very useful is given in the following theorem:
THEOREM: If a function f (x) assumes values of opposite signs at the end points of interval
𝐼 = [𝑎, 𝑏], i.e. if 𝑓 (𝑎) 𝑓 (𝑏) < 0 then the interval 𝐼 will contain at least one root of the equation
𝑓 (𝑥) = 0.
• We look at several techniques for solving 𝑓 𝑥 = 0.

2.1 BISECTION METHOD or INTERVAL-HALVING METHOD.

• The first technique is called the Bisection, or Interval-Halving method. The method assumes
that f is a continuous function defined on the interval [a, b], with f (a) and f (b) having
opposite signs. The method calls for a repeated halving (or bisecting) of subintervals of [a, b]
and, at each step, locating the half containing the root p.
2.1 BISECTION METHOD.
• The Bisection algorithm may be described as:
1. BEGIN
2. set 𝒊 = 𝟏
3. interval endpoint set 𝒂𝒊 = 𝒂 and set 𝒃𝒊 = 𝒃
4. REPEAT
𝒂𝒊 +𝒃𝒊
5. the midpoint of [𝑎𝑖 , 𝑏𝑖 ] set 𝒑𝒊 = 𝟐
6. If f (𝑝𝑖 ) = 0, then set p = 𝒑𝒊 and goto Step 11
7. If f (𝑝𝑖 ) and f (𝑎𝑖 ) have the same sign, set 𝒂𝒊+𝟏 =𝒑𝒊 and 𝒃𝒊+𝟏 =𝒃𝒊 and goto Step 9
8. If f (𝑝𝑖 ) and f (𝑏𝑖 ) have the same sign, set 𝒂𝒊+𝟏 =𝒂𝒊 and 𝒃𝒊+𝟏 =𝒑𝒊
9. set 𝒊 = 𝒊 + 𝟏
10. UNTIL satisfied
11. STOP
2.1 BISECTION METHOD.
• Error in approximated root

• Hence starting with interval [𝑎, 𝑏] and making n iterations, the approximate root 𝒑𝒏 has:
𝑏−𝑎
𝑎𝑏𝑠𝑜𝑙𝑢𝑡𝑒 𝑒𝑟𝑟𝑜𝑟 ≤ 𝑛
2
2.1 BISECTION METHOD

• From the 5th iteration 𝑎𝑛 , 𝑏𝑛 , and 𝑝𝑛 are equal up to two decimal places.

NB: All the calculation were carried out in radians mode not in degrees mode.
2.1 BISECTION METHOD
•

𝑎𝑏𝑠𝑜𝑙𝑢𝑡𝑒 𝑒𝑟𝑟𝑜𝑟𝑜𝑓 𝑝9 𝑖𝑠 ≤ ? ?

& 𝑎𝑏𝑠𝑜𝑙𝑢𝑡𝑒 𝑒𝑟𝑟𝑜𝑟 𝑜𝑓 𝑝3 𝑖𝑠 ≤? ?

2.2 NEWTON'S METHOD
• Newton's method (also called Newton-Raphson method) is one of the most widely used
methods of solving equations. It can be used to determine the roots of
𝑓 𝑥 =0 (eq1)
provided function 𝑓 𝑥 is continuous and has a first derivative.
1st DERIVATION: This method may be derived using the Taylor’s series
expansion about a point 𝑥0 in the vicinity of the exact root 𝑥.
′ 𝑥−𝑥0 2 ′′
𝑓 𝑥 = 𝑓 𝑥0 + 𝑥 − 𝑥0 𝑓 𝑥0 + 𝑔 𝑥0 + ⋯ = 0 (eq2)
2!
Since 𝑥0 in the vicinity of the exact root 𝑥, i.e. 𝑥 − 𝑥0 ≅ 0, we can neglect term involving
higher powers of 𝑥 − 𝑥0 . Thus eq2 takes the form:
𝑓 𝑥0 + 𝑥 − 𝑥0 𝑓 ′ 𝑥0 ≅ 0 (eq3)
The value of 𝑥 is obtained in eq3 by making 𝑥 the subject of the formula, in which case we get:
𝑓 𝑥0
𝑥 = 𝑥0 −
𝑓′ 𝑥0
Which suggests an iterative scheme of the form:
𝒇 𝒙𝒏
𝒙𝒏+𝟏 = 𝒙𝒏 − (eq4)
𝒇 ′ 𝒙𝒏
2.2 NEWTON'S METHOD
• Newton's method (also called Newton-Raphson method) is one of the most widely used methods
of solving equations. It can be used to determine the roots of 𝑓 𝑥 = 0 provided function
𝑓 𝑥 is continuous and has a first derivative.
• The algorithm for Newton’s method may simply be described as:
1. BEGIN
2. make initial guess 𝒙𝟎
3. set 𝒊 = 𝟎
4. REPEAT
𝒇(𝒙 )
5. compute 𝒙𝒊+𝟏 = 𝒙𝒊 − 𝒇′(𝒙𝒊 )
𝒊
6. set 𝒊 = 𝒊 + 𝟏
7. UNTIL satisfied
8. STOP

1st DERIVATION: This method may be derived using the Taylor’s series expansion
about a point 𝑥0 in the vicinity of the exact root 𝑥.
𝑥−𝑥0 2 ′′
𝑓 𝑥 = 𝑓 𝑥0 + 𝑥 − 𝑥0 𝑓 ′ 𝑥0 + 𝑔 𝑥0 + ⋯ = 0 (eq1)
2!
Since 𝑥0 in the vicinity of the exact root 𝑥, i.e. 𝑥 − 𝑥0 ≅ 0, we can neglect term involving higher
powers of 𝑥 − 𝑥0 . Thus eq1 takes the form:
𝑓 𝑥0 + 𝑥 − 𝑥0 𝑓 ′ 𝑥0 ≅ 0 (eq2)
The value of 𝑥 is obtained in eq2 by making 𝑥 the subject of the formula, in which case we get:
𝑓 𝑥0
𝑥 = 𝑥0 −
𝑓 ′ 𝑥0
Which suggests an iterative scheme of the form:
𝒇 𝒙𝒏
𝒙𝒏+𝟏 = 𝒙𝒏 − (eq4)
𝒇′ 𝒙𝒏
2.2 NEWTON'S METHOD
2nd DERIVATION: An alternative derivation of Newton’s method makes use of the graphical
∆𝑦
method. Recall that the slope 𝑚 of a straight line is given by: 𝑚 = and the slope of function
∆𝑥
𝑓 at 𝑥0 is 𝑓′(𝑥0 ). Now consider the figure on the right:
We have that
∆𝑦 = 𝑓 𝑥0 − 0 = 𝑓 𝑥0 , and ∆𝑥 = 𝑥0 − 𝑥1
𝑓 𝑥0
Therefore 𝑓′(𝑥0 ) = 𝑚 =
𝑥0 −𝑥1
making 𝑥1 the subject of the formula in above equation
𝑓 𝑥0
gives 𝑥1 = 𝑥0 −
𝑓′(𝑥0 )
𝒇 𝒙𝒏
Which suggests an iterative scheme of the form: 𝒙𝒏+𝟏 = 𝒙𝒏 −
𝒇′ 𝒙𝒏
2.2 NEWTON'S METHOD - Examples
•
2.2 NEWTON'S METHOD - Examples
• Example 2
Compute a zero of the function 𝑓 𝑥 = 3𝑥 + sin 𝑥 − 𝑒 𝑥 .
• Solution

• Accurate to 3 decimal places.

2.2 NEWTON'S METHOD- Examples
•
2.2 NEWTON'S METHOD- Examples
2.2 NEWTON'S METHOD- Exercises 2.2
2.3 FIXED-POINT ITERATION
• The problem of finding roots of 𝑓 𝑥 = 0 is closely related to the problem of finding fixed
points of 𝑥 = 𝑔 𝑥 . For a given root finding problem, we can rearrange 𝑓 𝑥 into an equivalent fixed
point problem 𝑥 = 𝑔 𝑥 so that if 𝑟 is a solution r = 𝑔 𝑟 it also a solution of 𝑓 𝑟 = 0

• Example: 𝑓 𝑥 = 𝑥 2 − 2𝑥 − 3 = 0 may be rewritten as

1 3
𝑥 = 𝑔1 𝑥 = 2𝑥 + 3, 𝑥 = 𝑔2 𝑥 = (𝑥 2 − 3), 𝑥 = 𝑔3 𝑥 =
2 𝑥−2
2
𝑥 = 𝑔4 𝑥 = 𝑥 − 𝑥 − 3, 𝑥 = 𝑔5 𝑥 = − 2𝑥 + 3, etc.
You can construct infinitely many 𝑔(𝑥)s from a given function 𝑓 𝑥 .
One can easily verify that above equation 𝑓 𝑥 = 0 has roots 𝑥 = −1 and 𝑥 = 3.

• Under suitable conditions (see theorem below), the iterative formula 𝒙𝒏+𝟏 = 𝒈 𝒙𝒏
converges to the fixed point r , which is a root of 𝒇 𝒙 = 𝟎.

THEOREM:
2.3 FIXED-POINT ITERATION
• Hence to guarantee convergence:
1. Both 𝑔 𝑥 and 𝑔′ 𝑥 must be continuous on an interval 𝐼 that includes the root 𝑟,
2. |𝑔′ 𝑥 | < 1 for all 𝑥 ∈ 𝐼, and
3. 𝑥0 must be chosen so that 𝑥0 ∈ 𝐼.

Iteration using 𝑔1 𝑥 = 2𝑥 + 3 with 𝑥0 = 4

CONDITION TO GUARANTEE CONVERGENCE

3
1.(a) 𝑔1 𝑥 = 2𝑥 + 3 continuous for 𝑥 ≥ −
2
1 3
1.(b) 𝑔′1 𝑥 = continuous for 𝑥 > −
2𝑥+3 2
3
Thus 𝑔1 𝑥 and 𝑔′1 𝑥 both cont. for 𝑥 > −
2
1
2. |𝑔′ 𝑥 | = < 1 for 2𝑥 + 3 > 1 i.e. for 𝑥 > −1
2𝑥+3
3. 𝑥0 = 4
Thus for interval 𝐼 = (−1, +∞) all three conditions
are satisfied.
Thus convergence was guaranteed.
2.3 FIXED-POINT ITERATION
• Hence to guarantee convergence:
1. Both 𝑔 𝑥 and 𝑔′ 𝑥 must be continuous on an interval 𝐼 that includes the root 𝑟,
2. |𝑔′ 𝑥 | < 1 for all 𝑥 ∈ 𝐼, and
3. 𝑥0 must be chosen so that 𝑥0 ∈ 𝐼.

1
Iteration using 𝑔2 𝑥 = (𝑥 2 − 3) with 𝑥0 = 4
2

CONDITION TO GUARANTEE CONVERGENCE

1
1.(a) 𝑔1 𝑥 = (𝑥 2 − 3) continuous for all 𝑥
2
1.(b) 𝑔′1 𝑥 = 𝑥 continuous all 𝑥
Thus 𝑔1 𝑥 and 𝑔′1 𝑥 both cont. forall 𝑥
2. |𝑔′ (𝑥)| = |𝑥| < 1 for −𝟏 < 𝑥 < 𝟏
3. 𝑥0 = 4
Conditions 2 and 3 not satisfied. There is no root in
(-1, 1) and 𝑥0 = 4 is outside this interval.
Thus convergence is NOT guaranteed.
2.3 FIXED-POINT ITERATION - Exercises

• Show that convergence is guaranteed in interval [1, 1.5] in (c)

• Show that convergence is guaranteed in interval [1, 2] in (d)
2.4 FIXED-POINT ITERATION
2.4 SECANT METHOD
2.4 SECANT METHOD - Exercises