A subspace of is an invariant subspace for if , that is, if implies .
Here are some examples of invariant subspaces.
- and are trivially invariant subspaces of any .
- The null space is an invariant subspace of because implies .
- If is an eigenvector of then is a -dimensional invariant subspace, since , where is the eigenvalue corresponding to .
-
The matrix
has a one-dimensional invariant subspace and a two-dimensional invariant subspace , where denotes the th column of the identity matrix.
Let be linearly independent vectors. Then is an invariant subspace of if and only if for . Writing , this condition can be expressed as
for some .
If in (1) then with square and nonsingular, so , that is, and are similar.
Eigenvalue Relations
We denote by the spectrum (set of eigenvalues) of and by the pseudoinverse of .
Theorem.
Let and suppose that (1) holds for some full-rank and . Then . Furthermore, if is an eigenpair of with then is an eigenpair of .
Proof. If is an eigenpair of then , and since the columns of are independent, so is an eigenpair of .
If is an eigenpair of with then for some , and , since being full rank implies that . Hence
Multiplying on the left by gives , so is an eigenpair of .
Block Triangularization
Assuming that in (1) has full rank we can choose so that is nonsingular. Let and note that implies and . We have
which is block upper triangular. This construction is used in the proof of the Schur decomposition with , an eigenvector of unit -norm, and chosen to be unitary.
The Schur Decomposition
Suppose has the Schur decomposition , where is unitary and is upper triangular. Then and writing , where is , and
where is , we have . Hence is an invariant subspace of corresponding to the eigenvalues of that appear on the diagonal of . Since can take any value from to , the Schur decomposition provides a nested sequence of invariant subspaces of .
Notes and References
Many books on numerical linear algebra or matrix analysis contain material on invariant subspaces, for example
- David S. Watkins. Fundamentals of Matrix Computations Third edition, Wiley, New York, USA, 2010.
The ultimate reference is perhaps the book by Gohberg, Lancaster, and Rodman, which has an accessible introduction but is mostly at the graduate textbook or research monograph level.
- Israel Gohberg, Peter Lancaster, and Leiba Rodman, Invariant Subspaces of Matrices with Applications, Society for Industrial and Applied Mathematics, Philadelphia, PA, USA, 2006 (unabridged republication of book first published by Wiley in 1986).
Related Blog Posts
This article is part of the “What Is” series, available from https://nhigham.com/index-of-what-is-articles/ and in PDF form from the GitHub repository https://github.com/higham/what-is.