Random Polynomials and Approximate Zeros of Newton’s Method
- 1 December 1990
- journal article
- Published by Society for Industrial & Applied Mathematics (SIAM) in SIAM Journal on Computing
- Vol. 19 (6) , 1068-1099
- https://doi.org/10.1137/0219075
Abstract
In this paper the authors study the size of the set of “approximate zeros” for Newton’s method, for a randomly chosen polynomial over certain distributions. For a degree d monic polynomial with coefficients chosen uniformly and independently in the unit ball, the results of this paper show, for example, that the set of approximate zeros is at least $Cd^{(-1.5-\varepsilon)}$ for any positive $\varepsilon$, with C depending only on $\varepsilon$.
Keywords
This publication has 7 references indexed in Scilit:
- On the worst-case arithmetic complexity of approximating zeros of polynomialsJournal of Complexity, 1987
- Computational Complexity: On the Geometry of Polynomials and a Theory of Cost: IISIAM Journal on Computing, 1986
- Newton’s Method Estimates from Data at One PointPublished by Springer Nature ,1986
- Computational complexity. On the geometry of polynomials and a theory of cost. IAnnales Scientifiques de lʼÉcole Normale Supérieure, 1985
- The fundamental theorem of algebra and complexity theoryBulletin of the American Mathematical Society, 1981
- Optimal Error Bounds for the Newton–Kantorovich TheoremSIAM Journal on Numerical Analysis, 1974
- On the Distribution of Roots of PolynomialsAnnals of Mathematics, 1950