Alphabeta Math
ExampleConstruction: AI-adaptedVerification: AI-generatedPipeline-generatedprecheck passjudge pass (gpt-6-luna)
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.

Lanczos handles repeated top eigenvalues, absent top components, and exact breakdown

Example

Check four branches of Lanczos largest Ritz values obey a Chebyshev start-angle bound, including degenerate branches: (i) A=5I1 and b=1; (ii) A=diag⁡(5,5,1) with b=(1,0,1)T/2; (iii) A=diag⁡(5,2,1) with b=e2; and (iv) the same A with b=(e1+e2)/2. Verify the Rayleigh–Ritz values, top-eigenspace components, and exact invariant-subspace breakdown dimensions.

Facts & Assumptions

Given: The matrices, vectors, Krylov spaces, and Rayleigh–Ritz values in the Example.

[F1]

Lanczos largest Ritz values obey a Chebyshev start-angle bound, including degenerate branches: if the top eigenvalue has multiplicity r=n, every Rayleigh–Ritz value is λ1; if Pb=0, every θm≤λr+1; if the unwanted interval collapses and Pb≠0, then θm=λ1 for m≥2; and exact breakdown at dimension d with Pb≠0 gives θd=λ1.

[F2]

The Krylov subspace Km(A,b)=span⁡{b,Ab,…,Am−1b}: Km(A,b)=span⁡{b,Ab,…,Am−1b}.

[F3]

The Arnoldi process for building an orthonormal basis of a Krylov subspace: exact breakdown at a column means its orthogonalized residual is zero.

[F4]

The Lanczos process as Arnoldi specialized to a Hermitian matrix, with the usual real-subdiagonal phase convention: for a real symmetric matrix, exact Lanczos is the corresponding exact Arnoldi process specialized to the Hermitian three-term recurrence.

Verification

1.1givenF1F2F3F4algebra

In case (i), take b=1. Then P=I, Pb=b≠0, and K1=R is invariant under A=5I1. The first Arnoldi residual is zero, so Lanczos breaks down at d=1; every nonzero vector has Rayleigh quotient 5, giving θm=5 for all m≥1 by [F1]. In case (ii), P projects onto span⁡{e1,e2}, so Pb=e1/2≠0. Here Ab=(5,0,1)T/2 and Ab−b=(4,0,0)T/2, so K2(A,b)=span⁡{e1,e3}, an invariant subspace. The first Arnoldi residual is nonzero, while Aq2∈K2=span⁡{q1,q2} makes the second residual zero; hence Lanczos breaks down at d=2. Since e1∈K2 has Rayleigh quotient 5, the maximum is θ2=5, as [F1] predicts for the collapsed interval branch.

1.2givenF1F2F3F4algebra

In case (iii), P projects onto e1, so Pb=0; Ae2=2e2 gives Km(A,e2)=span⁡{e2} for every m, and therefore θm=2=λr+1, in agreement with [F1]. At the first Arnoldi column, Ae2 already lies in K1=span⁡{e2}, so subtracting its projection leaves zero residual. Thus [F3]–[F4] give exact Lanczos breakdown at d=1 without a top component.

2.1givenF1F2F3F4algebra∎

In case (iv), b and Ab=(5e1+2e2)/2 are independent, so K2=span⁡{e1,e2}; this space is invariant under A. At the second Arnoldi column, Aq2∈K2=span⁡{q1,q2}, so subtracting its two orthogonal projections leaves zero residual. Thus [F3]–[F4] give exact Lanczos breakdown at d=2. Since Pb=e1/2≠0, [F1] gives θ2=5; directly, Ab−2b=3e1/2∈K2, so e1∈K2 and its Rayleigh quotient is 5.

Depends on

Used by

Nothing in the library uses this result yet.

Dependency tree · two levels

13 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