Iterative root approximation in \(p\)-adic numerical analysis (Q1049397)

The author studies various iterative algorithms related to Newton's method for approximating an isolated \(p\)-adic zero of a polynomial with \(p\)-adic coefficients. Asymptotic cost coefficients are listed for Newton's method, the secant method, Halley's method, two deformations of Halley's method and an inverse interpolation method. It is found with asymptotically efficient arithmetic that the secant method matches Newton's method and it is the best of the derivative-free iterations studied in this paper. On the other hand, Newton's method is asymptotically inferior to a variation of Halley's method and to the interpolation method, both of which use first derivatives. An explicit condition that guarantees convergence is found for each of the six methods.