Recursive algorithm without extra function evaluations for the Jacobian matrix of Viéta's polynomial system with applications (Q1765801)

The authors want to formulate an algorithm to compute all zeros of a complex polynomial simultaneously. To this end they formulate the problem for a polynomial of degree \(m\) in a complex space of dimension \(m\). The system of equations is based on Viéta's relations for the coefficients of the polynomial with its roots. The system of equations is evaluated recursively, and it is shown how the Jacobian of this system can be computed on the side without much extra cost. This gives the basis to use Newton's method. Of course Newton's method will require the solution of equations with this Jacobian, i. e. the action of its inverse. It is shown that the Jacobian is nonsingular for distinct roots. The paper ends with a small numerical example.