Numerical methods for roots of polynomials. II (Q1949539)

The problem of polynomial root-solving that starts with some crude but reasonably close initial approximations to the roots is practically important. It can be effective to specialize root-finders at this local stage by choosing them independently of the global methods that supply crude initial approximations. With this motivation, the author surveys a number of iterative root-finders and estimated their efficiency. This book is the second part in the series ``A handbook of methods for polynomial root-solving''. For the first part, see (2007; Zbl 1143.65002). The second part consists of Chapters 7--15. From Chapter 7 to Chapter 12, the author continues to discuss a lot of traditional and latest numerical methods and corresponding error estimates on the calculation of roots of polynomials. The main methods include 1. Bisection method. It starts from \(a_0\) and \(b_0\) such that \(P(a_0)P(b_0)<0\); 2. Secant method is based on the following \[ x_{n+1}=x_n-P(x_n)\frac{x_n-x_{n-1}}{P(x_n)-P(x_{n-1})}; \] 3. Successive approximation method: \(x_{n+1}=P(x_n)\); 4. Graeffe process. The roots of \(P(x)\) can be approximated by the coefficients of \(P^m(z)\) which is defined as \[ P^0(x)=P(x), \qquad P^{m+1}(x^2)=(-1)^nP^m(x)P^m(-x); \] 5. Integration method is based on the formula in complex analysis: \[ N=\frac{1}{2\pi i}\oint\limits_C\frac{P^\prime(z)}{P(z)}\,\text{d}z. \] The problem of polynomial root-finding is reduced to compute the generalized eigenvalues of a matrix. In Chapter 12, the author reviews the solutions of polynomials of degree \(\leq 4\) by radicals. In Chapter 13, the author gives various proofs of the fundamental theorem of algebra. In Chapter 14, the author discusses the numbers of roots of polynomials in some specific domains. This topic is very important and interesting since it is closely related to stability of differential equations and difference equations. In Chapter 15, the author reviews universal polynomial factorization: \[ \parallel P(x)-\prod\limits_{j=1}^n(x-z_j)\parallel\leq \epsilon \parallel P\parallel. \] It can approximate all zeros \(\{z_j^*\}\) of any polynomial \(P(x)\) with a fixed small errors: \[ |z_j-z_j^*|\leq \epsilon\qquad (|z_j|\geq 1)\quad\text{and}\quad \left|\frac{1}{z_j}-\frac{1}{z_j^*}\right|\leq \epsilon\qquad (|z_j|\leq 1). \] This book comprehensively covers traditional and latest methods on the calculation of roots of polynomials. The readers will benefit from this book greatly since these numerical methods in this book are accurate practical and have wide applications in control theory, information processing, statistics, etc. This book is well-written and accessible to general audiences with knowledge of calculus, Linear algebra, and Complex analysis.