Comparative Analysis of Root-Finding

Methods

Introduction

Purpose

Root-finding methods are numerical techniques used to solve nonlinear equations of the form

f (x) = 0.

Different methods differ in speed, reliability, derivative requirement, interval updating rule, and
computational cost.

Methods Compared

We compare the following four important methods:

1. Bisection Method

2. False Position Method (Regula Falsi)

3. Newton–Raphson Method

4. Secant Method

Method-wise Core Formulas and Rules

1. Bisection Method

Basic Idea
Bisection is a bracketing method. It repeatedly divides the interval into two equal halves and
selects the half where the root lies.

Main Formula

xl + x u
xr =
2

Condition for Starting

f (xl ) f (xu ) < 0

This ensures that a root lies inside the interval [xl , xu ].

1
 Update Rule
After computing xr ,
f (xl )f (xr ) < 0 ⇒ xu = xr

f (xl )f (xr ) > 0 ⇒ xl = xr

f (xr ) = 0 ⇒ exact root found

Approximate Relative Error

xnew − xold
εa = r r
× 100%
xnew
r

2. False Position Method

Basic Idea
False Position is also a bracketing method. Instead of taking the midpoint, it uses the
x-intercept of the straight line joining the two endpoints.

Main Formula

xl f (xu ) − xu f (xl )
xr =
f (xu ) − f (xl )

Condition for Starting

f (xl ) f (xu ) < 0

So the root is bracketed inside [xl , xu ].

Update Rule
After computing xr ,
f (xl )f (xr ) < 0 ⇒ xu = xr

f (xl )f (xr ) > 0 ⇒ xl = xr

f (xr ) = 0 ⇒ exact root found

Approximate Relative Error

xnew − xold
εa = r r
× 100%
xnew
r

2
3. Newton–Raphson Method

Basic Idea
Newton–Raphson is an open method. It uses the tangent line at the current estimate to
generate the next estimate.

Main Formula

f (xi )
xi+1 = xi −
f ′ (xi )

Starting Requirement

Need one initial guess x0 and derivative f ′ (x)

Update Rule

f (xi )
xi −→ xi+1 = xi −
f ′ (xi )
There is no interval update because this is not a bracketing method.

Approximate Relative Error

xi+1 − xi
εa = × 100%
xi+1

4. Secant Method

Basic Idea
Secant is an open method and a derivative-free variation of Newton–Raphson. It replaces the
derivative by the slope of the secant line through two recent points.

Main Formula

f (xi )(xi − xi−1 )

xi+1 = xi −
f (xi ) − f (xi−1 )

Starting Requirement

Need two initial guesses xi−1 , xi

Opposite signs are not compulsory.

3
 Update Rule

(xi−1 , xi ) −→ (xi , xi+1 )

That is, after each iteration,
xi−1 ← xi , xi ← xi+1 .

Approximate Relative Error

xi+1 − xi
εa = × 100%
xi+1

Common Stopping Criteria

Stopping Rules
A method is usually stopped when one of the following is satisfied:

|εa | ≤ εs

or
|f (xr )| ≤ tolerance
or a maximum number of iterations is reached.

Main Comparative Table

Aspect Bisection Method False Position Newton–Raphson Secant Method

(Regula Falsi)
Basic Idea Bisects interval and Uses straight line Uses tangent line at Similar to Newton but
selects the sub-interval between interval current point derivative-free; uses last two
where root lies endpoints to estimate points
root
Main Formula
xl + x u
xr =
2 xl f (xu ) − xu f (xl ) f (xi ) f (xi )(xi − xi−1 )
xr = xi+1 = xi − xi+1 = xi −
f (xu ) − f (xl ) f ′ (xi ) f (xi ) − f (xi−1 )

Initial Two guesses with Two guesses with Single guess + Two initial guesses
Requirement opposite signs opposite signs derivative
Condition for Good initial guess and Two starting values with
Use f ′ (x) ̸= 0 near root distinct function values
f (xl )f (xu ) < 0 f (xl )f (xu ) < 0

Update Rule Choose subinterval Choose subinterval Direct point update Shift pair:
using sign test using sign test only (xi−1 , xi ) → (xi , xi+1 )
Error Formula
xi+1 − xi
× 100%
xnew
r − xold
r xnew
r − xold
r xi+1 − xi xi+1
new
× 100% new
× 100% × 100%
xr xr xi+1

Guarantee of Always converges Converges but may Not guaranteed Not guaranteed
Convergence stagnate
Convergence Linear (slow) Linear, often faster Quadratic (fast) Super-linear
Rate than bisection

4
 Aspect Bisection Method False Position Newton–Raphson Secant Method
(Regula Falsi)
Iterations High Moderate Low Moderate
Needed
Function 1 per iteration 1 per iteration 2 per iteration (f, f ′ ) 2 per iteration
Evaluations
Derivative Not needed Not needed Required Not required
Requirement
Robustness Very robust Fairly robust, may slow Sensitive to guess, may Less robust than bisection
down diverge
Use Cases When guaranteed When bracketing is When derivative is When derivative is unavailable
convergence is essential desired but faster than available and good but faster convergence is
bisection initial guess exists desired
Accuracy Moderate Moderate, may stall High, fewer iterations Good, function-dependent

Update Interval or Update Point Conditions

Bracketing Methods: Bisection and False Position

Common Interval Update Logic

For both Bisection and False Position, after calculating the new estimate xr ,

f (xl )f (xr ) < 0 ⇒ root lies in [xl , xr ], xu = xr

f (xl )f (xr ) > 0 ⇒ root lies in [xr , xu ], xl = xr

f (xr ) = 0 ⇒ exact root

Open Methods: Newton–Raphson and Secant

Point Update Logic

Newton–Raphson and Secant do not preserve a bracket.

For Newton–Raphson:
xi → xi+1
For Secant:
(xi−1 , xi ) → (xi , xi+1 )
So they are faster, but they may move away from the true root.

Detailed Comparison by Feature

1. Safety and Reliability

Most Reliable
The Bisection Method is the safest because it always preserves the bracketing interval.

5
 Moderately Reliable

The False Position Method is also safe as a bracketing method, but it can become slow if one
endpoint hardly changes.

Less Reliable
Newton–Raphson and Secant are faster but not guaranteed to converge. They may:
• diverge,

• oscillate,

• jump to another root,

• fail when slope is zero or near zero.

2. Speed

Speed Ranking

Newton–Raphson > Secant > False Position > Bisection

in terms of speed of convergence, when they work properly.

3. Derivative Requirement

Important Distinction

• Bisection: no derivative

• False Position: no derivative

• Newton–Raphson: derivative required

• Secant: derivative not required

So Secant is very useful when Newton’s speed is desired but the derivative is hard to compute.

4. Cost Per Iteration

Cost Discussion
• Bisection and False Position are simple per iteration.

• Newton–Raphson may converge in fewer iterations, but each step requires derivative evaluation.

• Secant avoids derivative computation but still uses two recent function values.

6
Best Method Selection Guide

Quick Method Choice

Need guaranteed convergence? Use Bisection.

Need bracketing but usually faster than Bisection? Use False Position.

Have derivative and good initial guess? Use Newton–Raphson.

No derivative, but want faster than Bisection? Use Secant.

Advantages and Weaknesses at a Glance

Bisection Method

Advantages

• Always converges

• Very simple

• Highly stable

Weakness
Slow convergence.

False Position Method

Advantages

• Bracketing method

• Usually faster than bisection

• No derivative needed

Weakness
May stagnate because one endpoint can remain fixed.

Newton–Raphson Method

Advantages

• Very fast

• Quadratic convergence

• High accuracy with few iterations

7
 Weaknesses
• Needs derivative

• Sensitive to initial guess

• May diverge

Secant Method

Advantages

• No derivative needed

• Faster than bracketing methods in many cases

• Good alternative to Newton

Weakness
Less robust than bracketing methods.

Short Exam Notes

Very Important Points

1. Bisection and False Position are bracketing methods.

2. Newton–Raphson and Secant are open methods.

3. Bracketing methods require

f (xl )f (xu ) < 0.

4. Newton–Raphson requires derivative:

f (xi )
xi+1 = xi − .
f ′ (xi )

5. Secant avoids derivative:

f (xi )(xi − xi−1 )
xi+1 = xi − .
f (xi ) − f (xi−1 )

6. Bisection is most reliable.

7. Newton–Raphson is fastest when it converges.

8. False Position may stagnate.

9. Secant is a good compromise between speed and derivative-free computation.

8
Final Summary

One-box Summary

Bisection: safest and always convergent, but slow.

False Position: bracketing and often faster than bisection, but may stagnate.

Newton–Raphson: very fast and accurate, but needs derivative and good initial guess.

Secant: derivative-free alternative to Newton, moderately fast, but less robust.

Thus, the best method depends on reliability, derivative availability, and required speed.