A polynomial Jacobi-Davidson solver with support for non-monomial bases and deflation (Q2185867): Difference between revisions

Large-scale polynomial eigenvalue problems can be solved by Krylov methods operating on an equivalent linear eigenproblem (linearization) of size \(d\times n\), where \(d\) is the polynomial degree and \(n\) is the problem size, or by projection methods that keep the computation in the \(n\)-dimensional space. The Jacobi-Davidson technique belongs to the latter class of methods, and, since it is a preconditioned eigensolver, it may be competitive in cases where explicitly computing a matrix factorization is exceedingly expensive. However, a fully fledged implementation of a polynomial Jacobi-Davidson method has to consider several issues, including the deflation to compute more than one eigenpair, the use of non-monomial bases for the case of large degree polynomials, and the handling of complex eigenvalues when computing in real arithmetic. The authors discuss these aspects and present computational results of a parallel implementation in the SLEPc library.