On factorization of analytic functions and its verification (Q5926520): Difference between revisions

Let \(f\) denote an analytic function. Assume that \(f\) has a cluster of zeros \(z_1,\dots ,z_n\). The numerical computation of these zeros is extremely ill-conditioned. However, the location and multiplicity of a cluster of zeros of a polynomial is a stable phenomenon. Therefore the authors attempt to factor \(f(z)\) into \(f(z)=p(z)\cdot g(z)\), where \(p\) is a polynomial with \(p(z_k)=0\) and \(g(z_k)\neq 0\) for \(k=1,2,\dots ,n\). In the paper a method is described that generates a sequence of polynomials converging to \(p\). And an algorithm is presented that produces an interval-polynomial \(P\) with circular coefficients such that \(p(z)\in P(z)\). Some examples illustrate the numerical features of the presented algorithm.