Solving the sextic by iteration: A study in complex geometry and dynamics

Scott Crass
DOI: https://doi.org/10.48550/arXiv.math/9903111
1999-03-18
Dynamical Systems
Abstract:Recently, Peter Doyle and Curt McMullen devised an iterative solution to the fifth degree polynomial. At the method's core is a rational mapping of the Riemann sphere with the icosahedral symmetry of a general quintic. Moreover, this map posseses "reliable" dynamics: for almost any initial point, its trajectory converges to one of the periodic cycles that comprise an icosahedral orbit. This symmetry-breaking provides for a reliable or "generally-convergent" quintic-solving algorithm: with almost any fifth-degree equation, associate a rational map that has reliable dynamics and whose attractor consists of points from which one computes a root. An algorithm that solves the sixth-degree equation requires a dynamical system with the symmetry of the alternating group on six things. This group does not act on the Riemmann sphere, but does act on the complex projective plane--this is the Valentiner group. The present work exploits the resulting 2-dimensional geometry in finding a Valentiner-symmetric rational map whose elegant dynamics experimentally appear to be reliable in the above sense--transferred to the 2-dimensional setting. This map provides the central feature of a conjecturally-reliable sextic-solving procedure analogous to that employed in the quintic case. The paper culminates in an explicit description of the algorithm.
What problem does this paper attempt to address?