Documentation

Complexity.Computability.PolyTime.BooleanAnalysis.BoundedIncidenceStepCert

Native Step and Measurement-Payload Certificates for Bounded-Incidence Branch Evaluation #

This file establishes polynomial-time stage certificates and uniform ComputableInO / PolyBounded-workspace bounds for the inner group-cube and two-pass measurement-fiber payload evaluators of the bounded-incidence $\mathbb{F}_2$ polynomial-system branch algorithm of OpenAI, The Exponential Time Hypothesis is False (§3.3, Lemma 3.2, p. 23; §§5.3–5.4, Proposition 5.3, pp. 37–38; Appendix B.2–B.3, pp. 48–50).

Specifically, we define explicit bit-list encodings for the closed stage contexts GroupEvalCtx, MeasStepCtx, and SubsetStepCtx, certify validBranchEvalParams, validDesignShape, and evalGroupFactorJet in polynomial time, and compose BooleanMeasurements.searchCertificate (Pass 1) with CommonCountingAccum.computableInO_streamGroupJetSum and CommonCountingAccum.computableInPolyTime_prodJet (Pass 2) across the $r_u$ variable groups to certify the per-measurement-word payload evalMeasPayload in $O\bigl((r_u (2^{d_{\max}} + 1) + 1) \cdot \operatorname{poly}(|\mathrm{input}|)\bigr)$ time and globally polynomial workspace.

Main definitions #

Main results #

References #

1. Stage Encodings and Polynomial-Time Step Helpers #

Bit-list encoding for GroupEvalCtx, encoding (n', J, ℓ, k') in unary, (node, dDen) and measurement vectors in signed binary, bit masks via encBits, and precollected group equations with replacement-row tables as nested lists.

Instances For

    Bit-list encoding for MeasStepCtx, encoding (n', r_u, m_rows) and (J, ℓ, k') in unary together with node parameters, per-group variable/measurement slices, subset bit masks, and precollected group equations.

    Instances For

      Bit-list encoding for SubsetStepCtx, encoding (n', r_u, m_rows) and (J, ℓ, k') in unary, (c_meas, D_cap) as (encUnary, encNat), and the fixed per-node group and equation tables.

      Instances For

        validDesignShape is computable in polynomial time when (n', r_u, m_eq) are given in unary and data is encoded via encReplacementDesignData.

        2. Bounded-Range Loop Combinators and Measurement-Payload Certificate #

        Maximum unary-clamped variable-group dimension min (2 * n') |V_{t,g}| across g < r_u at a fixed interpolation node t.

        Instances For

          Maximum unary-clamped variable-group dimension across all nodes t < Q_u and groups g < r_u.

          Instances For

            Standalone per-measurement-word two-pass payload: slices word across the r_u variable groups, searches each unrestricted group fiber via BooleanMeasurements.search (Pass 1), and when every fiber is nonempty streams restricted local-cube factor-jet sums via CommonCountingAccum.streamGroupJetSum and returns some (prodJet J ℓ groupJets) (Pass 2).

            Instances For
              theorem Cslib.Complexity.SubexponentialSAT.SparsePolyCounter.measurementStepBody_eq (inp : ((ℕ × ℕ) × List ℤ) × MeasStepCtx × JetRows) :
              measurementStepBody inp = (evalMeasPayload (inp.1.2, inp.2.1)).elim inp.2 fun (y : JetRows) => (inp.2.1, addJet inp.2.1.2.1.1 inp.2.1.2.1.2.1 inp.2.2 y)

              measurementStepBody factors into evalMeasPayload followed by optional addJet accumulation.