Reliable Krylov-based Algorithms for Matrix Null Space

Eberly, Wayne

Reliable Krylov-based Algorithms for Matrix Null Space

dc.contributor.author	Eberly, Wayne	eng
dc.date.accessioned	2008-02-26T20:31:16Z
dc.date.available	2008-02-26T20:31:16Z
dc.date.computerscience	2004-12-20	eng
dc.date.issued	2004-12-20	eng
dc.description.abstract	Krylov-based algorithms have recently been used, in combination with other methods, to solve systems of linear equations and to perform related matrix computations over finite fields. For example, large and sparse systems of linear equations over F 2 are formed during the use of the number field sieve for integer factorization, and elements of the null space of these systems are sampled. Block Lanczos algorithms have been used to perform this computation with considerable success. However, the algorithms that are currently in use do not appear to be reliable in the worst case. This report presents a block Lanczos algorithm that is somewhat simpler than block algorithms that are presently in use and provably reliable for computations over large fields. This can be implemented, using a field extension, in order to produce several uniformly and independently selected elements from the null space at once. The amortized cost to produce each vector closely matches the cost to generate such a vector with the methods currently in use. An algorithm is also given to compute the rank of a matrix A E F m x n over a small finite field F. The expected number of matrix-vector products by A or A t used by this algorithm is in O (r), where r is the rank of A. The expected number of additional field operations used by this algorithm is within a polylog factor of r (n+m), and the expected storage space is within a polylog factor of n + m. This is asymptotically more efficient than existing black box algorithms to compute the rank of a matrix over a small field, assuming that the cost of matrix-vector products dominates the cost of other operations.	eng
dc.description.notes	We are currently acquiring citations for the work deposited into this collection. We recognize the distribution rights of this item may have been assigned to another entity, other than the author(s) of the work.If you can provide the citation for this work or you think you own the distribution rights to this work please contact the Institutional Repository Administrator at digitize@ucalgary.ca	eng
dc.identifier.department	2004-749-14	eng
dc.identifier.doi	http://dx.doi.org/10.11575/PRISM/30592
dc.identifier.uri	http://hdl.handle.net/1880/45469
dc.language.iso	Eng	eng
dc.publisher.corporate	University of Calgary	eng
dc.publisher.faculty	Science	eng
dc.subject	Computer Science	eng
dc.title	Reliable Krylov-based Algorithms for Matrix Null Space	eng
dc.type	unknown
thesis.degree.discipline	Computer Science	eng