How statement and proof provenance work
The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.
- Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
- AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
- AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.
These labels describe origin, not correctness: citations and verification chips remain separate evidence.
An exact Francis step terminates with unitary similarity and its stated status
Statement
For every valid input of Exact finite Francis single- or double-shift Hessenberg QR step, and every allowed choice of its reflectors, the procedure terminates. If , it returns , and zero-shift-vector. Otherwise it returns completed, with upper Hessenberg, unitary, and , where and . The bulge-chase-invariant-failure branch is unreachable for valid exact inputs. These claims include real exceptional quadratic descriptors and make no convergence assertion.
Facts & Assumptions
Given: An exact upper-Hessenberg , a valid shift descriptor, and exact arithmetic.
Upper Hessenberg matrices and real symmetric tridiagonal matrices: entries below the first subdiagonal of vanish.
Householder reflectors and Givens transformations are unitary and can annihilate prescribed entries: a Householder reflector is unitary and maps each nonzero active vector to a multiple of the first coordinate vector.
The real Francis double-shift polynomial is real and annihilates its trailing block: the real double-shift polynomial from a trailing block has real coefficients and factors over as its two roots.
A compact Francis bulge chase preserves Hessenberg form and the initial direction: length-2/3 windows restore the Hessenberg pattern and preserve the initial direction when every chosen factor aligns its active vector. The algorithm uses the identity only for zero vectors and otherwise chooses an aligning reflector.
Proof
If is upper Hessenberg, then is supported in rows ; applying once more makes supported in rows . Thus the defined is supported in its first entries for either shift descriptor.
If the initial vector is zero, the algorithm returns the identity step. Otherwise [F2] supplies . Each later active vector either is zero, in which case its identity branch is defined, or [F2] supplies . The loop has at most iterations, so the step terminates. In the ordinary real branch [F3] identifies the trace/determinant polynomial with the conjugate-pair shifts when the trailing block has negative discriminant.
Every update is the exact similarity . Hence the product gives by induction on the finite update list; every factor is unitary by [F2].
In the nonzero-vector branch, [F4] proves that the final matrix is upper Hessenberg and is proportional to , so the exact checks return completed. The zero-vector branch already returned its explicit status. Hence the invariant-failure branch is unreachable for valid exact inputs, and every returned status and similarity factor is well-defined.
Depends on
- Exact finite Francis single- or double-shift Hessenberg QR step
- Upper Hessenberg matrices and real symmetric tridiagonal matrices
- Unshifted QR iteration, shifted QR iteration, deflation, and the Wilkinson shift
- Householder reflectors and Givens transformations are unitary and can annihilate prescribed entries
- A compact Francis bulge chase preserves Hessenberg form and the initial direction
- The real Francis double-shift polynomial is real and annihilates its trailing block
Used by
Nothing in the library uses this result yet.
Cited to discharge well-definedness by Exact finite Francis single- or double-shift Hessenberg QR step.
Dependency tree · two levels
12 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.
Sources
- James Demmel, Applied Numerical Linear Algebra, §4.4.8 (standard reference, not scraped)
- David Bindel, CS 6210 Matrix Computations, Hessenberg QR iteration (standard reference, not scraped)