Alphabeta Math
ExampleConstruction: AI-generatedVerification: 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.

A two-vector LOBPCG block resolves a clustered pair in one finite Rayleigh-Ritz update

Example

Consider the mass-normalized vibration matrix A=diag⁡(1,101/100,5,9) with B=I and preconditioner T=I. The two wanted lowest modes are close, while the next mode is separated. Run the width-q=2 algorithm with residual tolerance τ=0 and budget Nmax⁡=1. Start with the orthonormal block x1=(99e1+e3)/10,x2=(99e2+e4)/10.

Verification

Given: The matrix, metric, preconditioner, and two starting vectors above.

[F1] Finite-budget block LOBPCG for a Hermitian-definite pencil with explicit rank-loss status: LOBPCG forms R=AX−BXΘ, W=TR, uses the trial span of X,W,D, and selects its q smallest Ritz values after Rayleigh-Ritz.

[F2] A successful LOBPCG Rayleigh-Ritz update cannot increase the selected Ritz values: on a successful update retaining X, each selected Ritz value cannot increase.

[F3] A principal-angle bound encloses the largest selected generalized Ritz value across a spectral gap: if the first q generalized eigenvalues are separated and sin⁡ϑ is the maximum principal angle, then θq(S)≤λq+(λn−λq)sin⁡2ϑ.

1.1givenF1algebra

The vectors are orthonormal and have disjoint coordinate supports. Their initial Rayleigh quotients are θ1=(99+5)/100=26/25 and θ2=((101/100)99+9)/100=10899/10000. Each xi has two nonzero coordinates associated with distinct diagonal entries of A, so ri=(A−θiI)xi≠0; also xi∗ri=xi∗Axi−θixi∗xi=0, and ri lies in the same two-coordinate plane as xi. Thus xi,ri are independent in that plane, and the disjoint planes make [X,R] rank four.

2.1givenF1F2step 1.1algebra

Since T=I, W=R and the first trial space is all of C4. Its Rayleigh-Ritz values are exactly 1,101/100,5,9, and the selected block is span⁡(e1,e2). The update is successful because the trial rank is four; the returned residuals are zero, so the algorithm returns converged. In particular, 1≤26/25 and 101/100≤10899/10000, as required by [F2].

3.1givenF3step 1.1algebra∎

For U=span⁡(e1,e2), every unit vector in S=span⁡(x1,x2) has squared norm 1/100 in U⊥, so sin⁡2ϑ=1/100. The strict gap 101/100<5 holds, and [F3] gives θ2(S)≤101/100+(9−101/100)/100=10899/10000, attained by the computed θ2. The block captures both clustered modes; a one-dimensional trial space cannot contain this two-dimensional invariant subspace.

Depends on

Used by

Nothing in the library uses this result yet.

Dependency tree · two levels

8 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