Extensions of Jordan bases for invariant subspaces of a matrix (Q802693)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Extensions of Jordan bases for invariant subspaces of a matrix |
scientific article |
Statements
Extensions of Jordan bases for invariant subspaces of a matrix (English)
0 references
1991
0 references
Let A be an \(n\times n\) complex matrix considered as a linear transformation \(C^ n\to C^ n\). A Jordan basis for an A-invariant subspace W is a basis of W which is the union of chains, i.e. of sets of nonzero vectors u, \((A-\lambda I)u,...,(A-\lambda I)^{k-1}u\) such that \((A-\lambda I)^ ku=0\). For \(W=C^ n\), a Jordan basis for W is called a Jordan basis for A. The existence of a Jordan basis for A is a well known fact. From a Jordan basis \(\{u_ i\), \((A-\lambda_ iI)u_ i,...,(A- \lambda_ iI)^{k_ i-1}u_ i\); \(i=1,...,t\}\) for A, certain A- invariant subspaces are seen immediately. They are spanned by the vectors of a Jordan basis for A \(:\;M=span\{(A-\lambda_ iI)^{m_ i}u_ i,\quad (A-\lambda_ iI)^{m_ i+1}u_ i,...,(A-\lambda_ iI)^{k_ i-1}u_ i| i=1,...,t\}\) and are called marked. Thus, a marked subspace is an A-invariant subspace M such that there is a Jordan basis for M which can be extended to a Jordan basis for A. If every Jordan basis for M can be extended to a Jordan basis for A, then M is said to be strongly marked. The authors characterize here those matrices A for which every invariant subspace is marked, resp. strongly marked, as the matrices with the property that, for every eigenvalue \(\lambda_ 0\) of A, the multiplicities of A corresponding to \(\lambda_ 0\), differ at most for 1, resp. are equal. Moreover, for an A-invariant subspace W, those Jordan bases B for W are characterized which can be extended to a Jordan basis for A. This characterization is given in terms of two properties: CP (for any x in B, \(Ax=0\) or Ax\(\neq 0\) and \(dpth(Ax)=dpth(x)+1)\) and DP (for any vector \(w=\sum_{b\in B}\alpha_ bb\) in W, \(dpth(w)=\min \{dpth(b):\) \(b\in B\), \(\alpha_ b\neq 0\})\).
0 references
strongly marked invariant subspaces
0 references
height and depth of a vector
0 references
constancy property
0 references
depth property
0 references
Jordan basis
0 references
characterization
0 references