Skip to content

MANE 3351

Lecture 11

Classroom Management

Agenda

  • Test 1 not graded
  • Lab 4 extended until end of day on Thursday October 9, 2025
  • Newton’s Method

Resources

Handouts

Newton's Method

Both the bisection and False Position methods were bracketing approaches to find roots that required two starting points that "bracketed" the root. Today's topic, Newton's method or Newton-Raphson method, require only one starting point. Newton's method also requires the knowledge of the derivative.

Geometric Inspiration

Brin (2020)1 demonstrate the geometric inspiration for Newton's method

Newton's Method

Newton's Method

  • The formula is simply
\[ x_{i+1}=x_i - \frac{f(x_i)}{f^\prime(x_i)} \]

New Example Problem

  • Consider a new function, \(f(x)=e^x+2^{-x}+2\cos(x)-6=0\)

Example Function for Newton's Method

First Derivative

  • Newton's Method requires the first derivative:
\[ \frac{d}{dx}\left[e^x+2^{-x}+2\cos(x)\right]=e^x-2^{-x}\ln(x)-2\sin(x) \]

Review of Derivatives

Pseudo-code

Brin (2020)1provides the following pseudo-code

Newton's Method pseducode

Vectorizing a Function

  • All Python functions considered so far have operated on scalars
  • Functions can be created to process a vector of values with a single call. This is called a vectorized function
  • Consider plotting the new function defined earlier using a vectorized function

Vectorized Function Code

import math
import numpy as np
import matplotlib.pyplot as plt
#
def f(x):
    return (math.exp(x)+2**-x+2*math.cos(x)-6)
# convert f(x) in a vectorized function that can be applied to array


vec_f=np.vectorize(f)

x=np.linspace(0,5,101)


#print(x)

# y=f(x)



f_x=vec_f(x)

#print(f_x)

fig, ax = plt.subplots()
ax.plot(x,f_x)
ax.axhline(y=0.0, color='r', linestyle='-')
ax.set(xlabel='x', ylabel='f(x)',
       title='New Function')
plt.show()

Newton Raphson Python Code

import math
def f(x):
    return (math.exp(x)+2**(-x)+2*math.cos(x)-6)
def f_prime(x):
    return(math.exp(x)-2.0**(-x)*math.log(2.0)-2*math.sin(x))
N=100
tol=0.0005
x_0=3.5
counter=0
for j in range(N+1):
    counter=counter+1
    x=x_0-f(x_0)/f_prime(x_0)
    print("x={}".format(x))
    if math.fabs(x-x_0)<tol:
        print("the root is {} with value {}, required {} steps".format(x,f(x),counter))
        break
    x_0=x
print("completed")

Convergence

  • Cheney and Kincaid (2004)3 study the performance of Newton's methods
  • Assumptions
  • \(f\) contains two continuous derivatives, \(f^\prime\) and \(f^{\prime\prime}\)
  • \(r\) is a simple root, \(f^\prime(r)\neq 0\)
  • If \(r\) is started sufficiently close to \(r\), converges quadratically to \(r\)

  • \(|r-x_{n+1}|\leq c|r-x_n|^2\)

  • In other words, \(x_{n+1}\) has approximately twice as many correct digits as \(x_n!\)

Other Comments

Kiusalaas (2013)2 provides the following introduction to the Newton-Raphson Method

The Newton-Raphson algorithm is the best known method of finding roots for a good reason: It is simple and fast. The only drawback of the methods is that it uses the derivative \(f^\prime(x)\) of the function as well as the function \(f(x)\) itself. Therefore, the Newton-Raphson method is usable only in problems where \(f^\prime(x)\) can be readily computed.

Importance of Good Starting Point

Cheney and Kincaid (2004)3, provide several illustrations of bad starting points and the problems that can occur.

Bad Starting Points

  1. Brin, L, (2020), Tea Time Numerical Analysis: Experiences in Mathematics, 3rd edition ↩↩

  2. Kiusalaas, J., (2013), Numerical Methods in Engineering with Python 3 ↩

  3. Cheny, W., and Kincaid, D., (2004), Numerical Mathematics and Computer, 5th edition ↩↩