Numerical mathematics, Newton's method

Click For Summary
SUMMARY

This discussion focuses on the derivation and application of Newton's method in numerical mathematics, specifically regarding the function G defined as G(x) = x + H(x)·F(x). The key points include the Taylor series expansion of G at a fixed point r, the relationship between the functions F and G, and the derivation of the Jacobian G'(r) at the point r. The condition G'(r) = 0 leads to the formula H(r) = -[F'(r)]^{-1}, which is crucial for implementing the vector Newton method.

PREREQUISITES
  • Understanding of Taylor series expansion in multivariable calculus
  • Familiarity with Jacobians and their role in numerical methods
  • Knowledge of fixed points and their significance in iterative methods
  • Basic concepts of Richardson extrapolation in numerical analysis
NEXT STEPS
  • Study the derivation of the Taylor series for multivariable functions
  • Learn about the application of Jacobians in Newton's method
  • Research the principles of fixed point theory in numerical mathematics
  • Explore Richardson extrapolation techniques and their error analysis
USEFUL FOR

Students and professionals in mathematics, particularly those studying numerical methods, algorithm developers, and anyone interested in enhancing their understanding of Newton's method and its applications in solving equations.

Wingeer
Messages
76
Reaction score
0

Homework Statement


"The goal of this task is to check that you understood the derivation of Newton’s method in the lecture.
1. Consider a smooth function G defined from \mathbb{R}^N to \mathbb{R}^N. Suppose it admits a fixed point r \in \mathbb{R}^N. Write down the Taylor development of this function at the point r.
2. Consider now a function F also defined from \mathbb{R}^N to \mathbb{R}^N. Define a function G as:
G(x) = x + H(x)\cdot F(x) (1)
where H(x) is now a matrix depending on x, and \cdot denotes the matrix vector multiplication. Check that if r is a root of F (i.e., F(r) = 0), then r is also a fixed point of G.
3. G'(r) denotes the Jacobian of G at the point r. Suppose that r is a root of F, and that G takes the form (1). Write down the condition G'(r) = 0 and deduce a formula for H(r).
4. What is a natural choice of the matrix H that leads to the vector Newton method?"

Homework Equations




The Attempt at a Solution


I was not attending the lecture, so I am not sure how this was derivated.
The Taylor series is: G(x_1, x_2, \cdots , x_n) \equiv G(\mathbf{x}) = G(r) + \sum_{j=1}^{n} r_j \frac{\partial G(r)}{\partial x_j} + O(r^2)
b) Trivial by insertion. I do however not see the relationship between the function given in b, and the one in a?
c) I suppose G'(r)=0=1 + H'(r)F(r) + H(r)F'(r) = 1 + H(r)F'(r) from which:
H(r)=-[F'(r)]^{-1}.
d) I am really not sure. I cannot see the whole picture here. And what's the point of the fixed point?

Thanks in advance.
 
Physics news on Phys.org
Partly as a bump, an partly another question related to numerical mathematics I post in my own thread.
"Consider the formula \phi(h)=\frac{f(x_0 +h) - f(x_0)}{h} Use Richardson extrapolation on the points h and h/2 in order to find a new formula. What should be its new order?"
What is meant by the order? And how do I use Richardson extrapolation? Am I not dependent on some error terms to use the method?
 

Similar threads

  • · Replies 1 ·
Replies
1
Views
2K
Replies
3
Views
2K
Replies
1
Views
1K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 3 ·
Replies
3
Views
1K
Replies
2
Views
1K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 3 ·
Replies
3
Views
1K
  • · Replies 3 ·
Replies
3
Views
2K