On Positive Semidefinite Matrices with Known Null Space (Q4785765)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1848584
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | On Positive Semidefinite Matrices with Known Null Space |
scientific article; zbMATH DE number 1848584 |
Statements
On Positive Semidefinite Matrices with Known Null Space (English)
0 references
5 January 2003
0 references
positive semidefinit matrices
0 references
Cholosky factorization
0 references
null space basis
0 references
consistent systems
0 references
constrained systems
0 references
generalized eigenvalue problems
0 references
algorithms
0 references
0.7687729001045227
0 references
0.762651264667511
0 references
0.7461575269699097
0 references
The authors investigate procedures to exploit the knowledge of an explicite basis of the nulll space of an \(n\times n\) real symmetric positive semidefinit matrix \(A\). Specifically, if \(Y\) is an \(n\times m\) matrix such that its colums span the null space of \(A\) then the knowledge that its last \(m\) rows are linearly indepenedent is exploited efficiently in the investigation of consistent systems of equations, constrained systems of equations, and generalized eigenvalue problems. Such knowledge is used to extract a priori a maximal positive semidefinite submatrix so that the order of the problem is reduced by the dimension of the null space of \(A\). It is observed that the backward error in the null space in the context of Cholosky factorization is small if the error in the Cholosky factor is `almost' orthogonal to the null space of \(A\). The factorization of \(A\) considered for the case of systems of equations and eigenvalue problems leads to interesting varieties of the original algorithms and also leads to more accurate results.
0 references