Alphabeta Math
TheoremStatement: Literature-sourcedProof: 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.

Separate stationary solve residual from generator error

Statement

Let Q be the true finite column generator, let ν>0 satisfy ν≥max⁡jrj, and put P=I+Q/ν. Suppose q=δ(Pm)<1 for some m≥1. Let Q~ be any computed matrix model on the same finite state set, let y∈Rn, and let x be a column probability vector with c=∥y−x∥1. Then, for the unique stationary probability πQ of Q,

∥y−πQ∥1≤c+mν(1−q)(∥Q~x∥1+∥(Q−Q~)x∥1).

The first residual is the solve residual for the computed model; the second is the model perturbation evaluated at the chosen candidate. If Q~x=0, the solve-residual term vanishes. Every term must be supplied as an exact value or a certified outward upper bound before the displayed expression is reported as a numerical certificate.

In particular, if P and y are rational and the exact branch of Correct and certify a finite-chain stationary candidate returns certified, its output x and c satisfy these conditions.

Facts & Assumptions

Given: Q, ν, P, m, q, y, a computed model Q~, a column probability vector x, and c=∥y−x∥1.

[F1]

Correct and certify a finite-chain stationary candidate: on exact-branch success for rational P, y, and m, x is a column probability vector and c=∥y−x∥1.

[F2]

Uniformized finite-generator stationary residual certificate: ∥y−πQ∥1≤c+m∥Qx∥1/(ν(1−q)) for the true generator Q and the chosen column probability vector x.

Proof

Given: Q, Q~, ν, P, q, m, y, x, and c under the stated hypotheses.

Proof technique: write the exact generator residual as the sum of the computed solve residual and the model discrepancy.

1.1givenalgebra

The exact generator residual decomposes as Qx=Q~x+(Q−Q~)x. The triangle inequality gives ∥Qx∥1≤∥Q~x∥1+∥(Q−Q~)x∥1.

2.1F1F2step 1.1algebra∎

By [F2], the generator certificate gives ∥y−πQ∥1≤c+m∥Qx∥1/(ν(1−q)). Substitute the estimate from step 1.1; since ν(1−q)>0, the result is the stated separate solve-residual and model-error bound. An outward upper bound for either norm preserves the inequality. If P and y are rational and the exact algorithm returns certified, [F1] supplies the required probability x and correction c.

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