 # Householder’s Root Finding Method of Linear and Nonlinear Equations

by admin in on June 14, 2019

Householder’s methods are a class of root-finding algorithms that are used for functions of one real variable with continuous derivatives up to some order d + 1. Each of these methods is characterized by the number d, which is known as the order of the method. The algorithm is iterative and has a rate of convergence of d + 1. These methods are named after the American mathematician Alston Scott Householder.

This Code is a Generalization method for Newton’s and Hally’s, furthermore, it is more accurate and faster than both, where by choosing methods’ order d = 1 ( this code will be Newton’s method) and by d = 2 ( will be Halley’s method) and for d >= 3 you will get very fast and accurate method than these.

### Example On Using This Code:

Input

d = 3;                     % Methods' Order {d = 1 (Newton's Method, d = 2 (Halley's)}
syms X;                    % Defining variable X for finding the derivations
fx = X^3-2*X-5;            % rewriting the equation we interested to solve
[Fpd,Fpd_1]=Fun(X,fx,d);   % calculating the d and d-1 derivatives of 1/f
err = 1e-9; Err = 1;       % err is the accuracy of the answer; Err > err

Output  The Root X = 2.0946, with Error of = 0, and # Iterations = 6  ### Contents

• Method
• Motivation
• First approach
• Second approach
• The methods of lower order
• Example
• Derivation
• Relation to Newton’s method
• References

### Method

Householder’s method is a numerical algorithm for solving the nonlinear equation f(x) = 0. In this case, the function f has to be a function of one real variable. The method consists of a sequence of iterations beginning with an initial guess x0.

If f is a d + 1 times continuously differentiable function and a is a zero of f but not of its derivative, then, in a neighborhood of a, the iterates xn satisfy: , for some This means that the iterates converge to the zero if the initial guess is sufficiently close, and that the convergence has order d + 1. Despite their order of convergence, these methods are not widely used because the gain in precision is not commensurate with the rise in effort for large d. The Ostrowski index expresses the error reduction in the number of function evaluations instead of the iteration count.

• For polynomials, the evaluation of the first d derivatives of f at xn using the Horner method has an effort of d + 1 polynomial evaluations. Since n(d + 1) evaluations over niterations give an error exponent of (d + 1)n, the exponent for one function evaluation is (d+1)^(1/(d+1)), numerically 1.41421.44221.41421.3797 for d = 1, 2, 3, 4, and falling after that.
• For general functions the derivative evaluation using the Taylor arithmetic of automatic differentiation requires the equivalent of (d + 1)(d + 2)/2 function evaluations. One function evaluation thus reduces the error by an exponent of 2^(1/3) = 1.2599 for Newton’s method, 3^(1/6) = 1.2009 for Halley’s method and falling towards 1 or linear convergence for the higher order methods.

### Motivation

#### First Approach

An approximate idea of Householder’s method derives from the geometric series. Let the real-valued, continuously differentiable function f(x) have a simple zero at x = a, that is a root f(a) = 0 of multiplicity one, which is equivalent to f'(a) =/= 0. Then 1/f(x) has a singularity at a, specifically a simple pole (also of multiplicity one), and close to a the behavior of 1/f(x) is dominated by 1/(x − a). Approximately, one gets Here f'(a) =/= 0 because a is a simple zero of f(x). The coefficient of degree d has the value C.a^(-d). Thus, one can now reconstruct the zero a by dividing the coefficient of degree d − 1 by the coefficient of degree d. Since this geometric series is an approximation to the Taylor expansion of 1/f(x), one can get estimates of the zero of f(x) – now without prior knowledge of the location of this zero – by dividing the corresponding coefficients of the Taylor expansion of 1/f(x) or, more generally, 1/f(b+x). From that one gets, for any integer d, and if the corresponding derivatives exist, #### Second Approach

Suppose x = a is a simple root. Then near x = a(1/f)(x) is a meromorphic function. Suppose we have the Taylor expansion: By König’s theorem, we have: This suggests that Householder’s iteration might be a good convergent iteration. The actual proof of the convergence is also based on this idea.

### The Methods of Lower Order

Householder’s method of order 1 is just Newton’s method, since: For Householder’s method of order 2 one gets Halley’s method, since the identities and result in In the last line, h_(n) = – f(x_(n)) / f'(x_(n)) is the update of the Newton iteration at the point x_(n). This line was added to demonstrate where the difference to the simple Newton’s method lies.

The third order method is obtained from the identity of the third order derivative of 1/f and has the formula and so on.

### Example

The first problem solved by Newton with the Newton-Raphson-Simpson method was the polynomial equation y^(3) – 2.y – 5 = 0. He observed that there should be a solution close to 2. Replacing y = x + 2 transforms the equation into .

The Taylor series of the reciprocal function starts with The result of applying Householder’s methods of various orders at x = 0 is also obtained by dividing neighboring coefficients of the latter power series. For the first orders one gets the following values after just one iteration step: For an example, in the case of the 3rd order, .

d x1
1 0.100000000000000000000000000000000
2 0.094339622641509433962264150943396
3 0.094558429973238180196253345227475
4 0.094551282051282051282051282051281
5 0.094551486538216154140615031261961
6 0.094551481438752142436492263099118
7 0.094551481543746895938379484125813
8 0.094551481542336756233561913325371
9 0.094551481542324837086869382419375
10 0.094551481542326678478801765822984

As one can see, there are a little bit more than d correct decimal places for each order d. The first one hundred digits of the correct solution are

0.0945514815423265914823865405793029638573061056282391803041285290453121899834836671462672817771577578.

Let’s calculate the x_(2), x_(3), x_(4) values for some lowest order,    And using following relations, 3rd order; x 1st (Newton) 2nd (Halley) 3rd order 4th order
x1 0.10000000000000000 0.09433962264150943 0.09455842997323818 0.09455128205128
x2 0.09456812110418521 0.09455148154016421 0.09455148154232659
x3 0.09455148169819930 0.09455148154232659 0.09455148154232659
x4 0.09455148154232659 0.09455148154232659 0.09455148154232659
x5 0.09455148154232659
x6 0.09455148154232659

### Derivation

An exact derivation of Householder’s methods starts from the Padé approximation of order d + 1 of the function, where the approximant with linear numerator is chosen. Once this has been achieved, the update for the next approximation results from computing the unique zero of the numerator. The Padé approximation has the form The rational function has a zero at h = – a_(0).

Just as the Taylor polynomial of degree d has d + 1 coefficients that depend on the function f, the Padé approximation also has d + 1 coefficients dependent on f and its derivatives. More precisely, in any Padé approximant, the degrees of the numerator and denominator polynomials have to add to the order of the approximant. Therefore, b_(d) = 0 has to hold.

One could determine the Padé approximant starting from the Taylor polynomial of f using Euclid’s algorithm. However, starting from the Taylor polynomial of 1/f is shorter and leads directly to the given formula. Since has to be equal to the inverse of the desired rational function, we get after multiplying with a_(0) + h in the power h^(d) the equation .

Now, solving the last equation for the zero h = – a_(0) of the numerator results in .

This implies the iteration formula .

### Relation to Newton’s Method

Householder’s method applied to the real-valued function f(x) is the same as Newton’s method applied to the function g(x): with In particular, d = 1 gives Newton’s method unmodified and d = 2 gives Halley’s method.

Share Now!

#### Release Information

• Price
:

\$8.99

• Released
:

June 14, 2019

• Last Updated
:

June 22, 2019