Eigenvalue computations in the context of data-sparse approximations of integral operators (Q455840)

Some iterative eigensolvers are used to compute a few eigenelements in the discretized formulation of an integral operator that appears in the radiative transfer equation in stellar atmospheres. The authors have implemented a SLEPc-based code for the H-matrix representation into the freely available software HLib, using the Krylov-Schur method. In the implementation of data-sparse approximation, the Taylor and Lagrange interpolation are used and the numerical experiments show the effectiveness of these interpolations in comparison with the sparse storage for increasing dimension.