Lars Karlsson
Office
UMIT Research Lab
Phone
+46 (0)90 786 70 24
Email
larsk@cs.umu.se
Address
Lars Karlsson
Dept. of Computing Science
Umeå University
SE-901 87 Umeå
Sweden
Teaching
Present
- Matrix Computations and Applications (HT11)
Past
Publications
- Lars Karlsson and Bo Kågström.
Parallel two-stage reduction to Hessenberg form using shared memory.
Report UMINF 10.14 (submitted to PMAA 2010), 2010.
- Lars Karlsson and Bo Kågström.
Efficient reduction from block Hessenberg form to Hessenberg form using shared memory.
Report UMINF 10.12 (submitted to PARA 2010), 2010.
- Bo Kågström, Lars Karlsson, and Daniel Kressner.
Computing Codimensions and Generic Canonical Forms for Generalized Matrix Products.
Report 2010-17, SAM, ETH Zurich, Switzerland (submitted to ELA), 2010.
- Fred Gustavson, Lars Karlsson, and Bo Kågström.
Parallel and Cache-Efficient In-Place Matrix Storage Format Conversion
ACM Transactions on Mathematical Software (submitted February 2010).
(Also published as Report UMINF 10.05.)
- Lars Karlsson.
Blocked and Scalable Matrix Computations --- Packed Cholesky, In-Place Transposition, and Two-Sided Transformations.
Licentiate Thesis, Dept. of Computing Science, Umeå University, Sweden, 2009.
Report UMINF 09.11, ISBN 978-91-7264-788-6.
- Lars Karlsson.
Blocked In-Place Transposition with Application to Storage Format Conversion.
Technical Report UMINF 09.01, Dept. of Computing Science, Umeå University, Sweden, 2009.
- Lars Karlsson and Bo Kågström.
A Framework for Dynamic Node-Scheduling of Two-Sided Blocked Matrix Computations.
In Proceedings of PARA 2008 (accepted), 2009.
- Fred Gustavson, Lars Karlsson, and Bo Kågström.
Distributed SBP Cholesky Factorization Algorithms with Near-Optimal Scheduling.
ACM Transactions on Mathematical Software, Volume 36, Number 2, pages 11:1-11:25, 2009.
(Also published as Report UMINF 07.19 and IBM Research Report RC24342.)
- Fred Gustavson, Lars Karlsson, and Bo Kågström.
Three Algorithms for Cholesky Factorization on Distributed Memory using Packed Storage.
In Applied Parallel Computing: State of the Art in Scientific Computing (PARA 2006), Lecture Notes in Computer Science, LNCS 4699, pages 550-559, Springer, 2007.
Software
In-Place Matrix Transposition and Matrix Storage Format Conversion
Software to efficiently transpose a matrix in-place or
convert to/from the standard column- and row-major matrix
storage formats and the four standard blocked formats.
- Source code (Fortran95 using OpenMP)
- The underlying theory is presented in [4]
- Collaborators: Fred Gustavson and Bo Kågström
Codimensions of Generalized Matrix Products
Software to compute the codimension of a generalized matrix product given
in canonical form.
Distributed SBP Cholesky Factorization
Prototype software to efficiently compute a dense Cholesky
factorization using the Distributed Square Block Packed
(Distributed SBP) storage format on a distributed memory
machine.
- Source code (C using MPI)
- The underlying theory is presented in [8]
- Collaborators: Fred Gustavson and Bo Kågström