Householder tridiagonalization reduces a real symmetric matrix by an orthogonal similarity to a real symmetric tridiagonal matrix :At column , a Householder reflection on coordinates eliminates entries below row . Apply the same reflection from the right, preserving symmetry. Since entries below the first subdiagonal become zero, symmetry also makes entries above the first superdiagonal zero. Earlier zeros are preserved by the restricted active coordinates.
The eigenvalues are unchanged by the similarity transformation; an eigenvector of gives the eigenvector of . Dense reduction costs arithmetic operations once, after which each structured QR algorithm step on the tridiagonal form is much cheaper than a new dense factorization. This is why eigensolvers separate reduction from iterative refinement of eigenvalues.
An orthogonal similarity preserves skew symmetry, so upper-Hessenberg reduction of a real skew-symmetric matrix is automatically tridiagonal with zero diagonal. For a Householder reflection, the scalar vanishes, eliminating the fourth term of the general similarity update. Storing only one triangle and using the displayed skew rank-two update costs scalar products, the same leading product count as optimized symmetric tridiagonalization; the simplification saves lower-order work.
Articles by others on the same topic
There are currently no matching articles.