A. I. Iliev,Kh. I. Semerdzhiev
Abstract:Iterative methods for the simultaneous determination of all roots of an equation are dis-cussed. The multiplicities of the roots are assumed to be known in advance. The methods are proved to have a cubical rate of convergence. Numerical examples are given.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to develop an improved Chebyshev Method for simultaneously determining all roots of a polynomial equation. Specifically, the objectives of the paper include:
1. **Simultaneously solve all roots**: Traditional methods can usually find only one root at a time, while the method proposed in this paper can find all roots of a polynomial equation simultaneously.
2. **Handle roots of any multiplicity**: Not limited to simple roots, this method can also handle roots with any given multiplicity.
3. **Increase the convergence speed**: The proposed method has a cubical rate of convergence, which means that as the number of iterations increases, the error will decrease rapidly.
4. **Reduce computational complexity**: Compared with the traditional Chebyshev - Markov method, the new method is more efficient and reduces the amount of computation, especially avoiding the calculation of high - order determinants in each iteration.
### Overview of the main content
#### 1. Introduction
The article reviews the development of classical numerical solutions in the past few decades and introduces several applications of the generalized Newton method, the Obrechkoff method, and the Chebyshev method. In particular, it mentions the application of the Chebyshev method in simultaneously solving all roots and points out its limitations, such as being only applicable to simple roots and having high computational complexity.
#### 2. Algebraic polynomials
For a given algebraic polynomial \(A(x) = a_n x^n + a_{n - 1}x^{n - 1}+\cdots+a_1x + a_0\), assume that all its roots are real numbers. A new iterative formula is proposed in the paper:
\[
x_i^{[k + 1]}=x_i^{[k]}-\frac{A(x_i^{[k]})}{A'(x_i^{[k]})+\frac{A''(x_i^{[k]})}{2}Q_i^{[k]}(x_i^{[k]})}
\]
where
\[
Q_i^{[k]}(x)=\prod_{\substack{j = 1\\j\neq i}}^m(x - x_j^{[k]})
\]
#### 3. Trigonometric polynomials
For a trigonometric polynomial \(T(x)=\sum_{k = 0}^n(a_k\cos(kx)+b_k\sin(kx))\), a similar iterative formula is given in the paper:
\[
x_i^{[k + 1]}=x_i^{[k]}-\frac{T(x_i^{[k]})}{T'(x_i^{[k]})+\frac{T''(x_i^{[k]})}{2}Q_i^{[k]}(x_i^{[k]})}
\]
where
\[
Q_i^{[k]}(x)=\prod_{\substack{j = 1\\j\neq i}}^m\left(\sin\left(\frac{x - x_j^{[k]}}{2}\right)\right)^{-2}
\]
#### 4. Exponential polynomials
For an exponential polynomial \(E(x)=\sum_{k = 0}^n(a_k\cosh(kx)+b_k\sinh(kx))\), a corresponding iterative formula is also given in the paper:
\[
x_i^{[k + 1]}=x_i^{[k]}-\frac{E(x_i^{[k]})}{E'(x_i^{[k]})+\frac{E''(x_i^{[k]})}{2}Q_i^{[k]}(x_i^{[k]})}
\]
where
\[
Q_i^{[k]}(x)=\prod_{\substack{j = 1\\j\neq i}}^m\left(\sinh\left(\frac{x - x_j^{[k]}}{2}\right)\right)^{-2}
\]
### Conclusion
Through the above methods, the paper proves that these iterative formulas have a cubical rate of convergence and demonstrates their effectiveness through numerical examples. This makes the new method more efficient and accurate in practical applications, especially suitable for the need to simultaneously solve multiple roots.