Roots of polynomials by Newton's iteration

An iterative method is described which finds all the roots of a square-free polynomial at once, using the original coefficients in each step (thus minimizing round off error). Since the iteration is basically Newton's, the usual convergence criteria and quadratic-convergence behavior obtain. Except for the facts about polynomial coefficients and contraction mappings, the paper is self-contained.Technical report DCS-TR-1