Learning Reduced Order Dynamics via Geometric Representations
Imran Nasim, Melanie Weber
SCML 2024
We present a numerical algorithm for the solution of a large number of shifted linear systems for which the system pencil is symmetric and definite and the shifts lie inside a given real interval. Extending an earlier method due to Meerbergen and Bai [SIAM J. Matrix Anal. Appl., 31 (2010), pp. 1642–1662], the algorithm uses a rational filter with poles at Chebyshev points to compute and deflate the components of the solution in the direction of eigenvectors of the system pencil corresponding to eigenvalues within the interval. It then solves the deflated systems for the remaining components using a Krylov subspace method with a preconditioner constructed by interpolating the factorizations at the filter poles. The algorithm parallelizes naturally. We demonstrate its effectiveness using matrix pencils from both model and real-world problems and discuss applications to frequency response analysis.
Imran Nasim, Melanie Weber
SCML 2024
Y.Y. Li, K.S. Leung, et al.
J Combin Optim
Hans Becker, Frank Schmidt, et al.
Photomask and Next-Generation Lithography Mask Technology 2004
Corneliu Constantinescu
SPIE Optical Engineering + Applications 2009