Fredagen den 24 maj kl 13.15-14.15 talar prof. Axel Ruhe, Chalmers, om The Rational Krylov algorithm for large nonsymmetric matrix eigenvalue problems
Abstract: The Rational Krylov algorithm is a generalization of shifted and inverted Arnoldi. In each run, Rational Krylov computes one sequence of orthogonal basis vectors, using factorizations of the matrix for several different shifts. This makes it possible to use information gathered while computing one eigenvalue to give a flying start in the iteration for the next. It is also possible to do several factorizations in parallel, and let vectors from all of them build up the orthogonal basis.
Results for some very ill
conditioned eigenvalue problems from applications will
be demonstrated.
Lokal: MAI:s seminarierum
Lars Eldén