Imports
import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.ReachableAlphabet import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Configuration import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.CircuitBuilder import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.CircuitBuilder.ConstantPool import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.CircuitBuilder.FiniteFamily import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.BundleCombinators import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.StackPrimitives import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.StackSemantics import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.FiniteLookup import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.StackCircuits import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.ControlCircuits import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.PrimitiveRowSemantics import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.Validity import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.ValidityIndices import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.ValidityBounds import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.Workspace import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.StatementCircuits import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.TransitionCircuits import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.BoundaryCircuits import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Tableau.Finishing import CLRSLean.Chapter_34.Section_34_4_NP_Completeness_Proofs.CookLevin.Circuitization

Cook--Levin infrastructure

The first layer extracts a finite program-support over-approximation of the alphabet used by any fixed bundled TM2 machine and proves that the invariant is preserved by finite execution. The configuration layer gives finite bounded row codes, exact two-way codecs, stuttering execution semantics, and a uniform height bound. The circuit-builder layer supplies proved fresh-wire allocation, a reusable exact two-gate Boolean constant pool, Boolean combinators, and exact-cost finite-family mux/equality kernels. The pure tableau layer fixes exact row coordinates and adds typed stack views, zero-gate stack replacement, exact-cost whole-row mux/equality, one-hot codecs, and fresh external-input layouts. It also gives pure fixed-width push, peek, and pop operations on raw Boolean stack bundles, with supported-head codecs, coordinate laws, and raw one-hot preservation under explicit capacity premises. The canonical semantics bridge proves exact list-level push, peek, and pop behavior and projects successful whole-row evaluation to every decoded machine stack. The wire layer lifts these operations to circuit builders: push and peek allocate no gates, positive-width pop allocates one OR, and capacity allocates one NOT, all with row-frame and list-semantic contracts. The generic finite-lookup layer adds exact-cost one-hot maps, pair maps, and Boolean predicate queries using only static finite preimages. The finite-control layer adds pool-backed state/status encodings, whole-row control replacement, canonical decoding inversion, and exact state/status update semantics. The primitive-row bridge proves that wire-level push and pop decode as complete dependent-stack updates and exposes pop's old head as an exact canonical one-hot family. The recursive statement compiler now covers all seven bundled statement constructors through finite state/symbol truth tables, proves exact stepAux semantics under explicit prefix capacity, and publishes exact and affine emitted-gate bounds. The local transition layer now widens the current public row, serially compiles every finite program label from that same workspace source, selects a complete row, narrows with an explicit fit bit, and compares the entire result to the next public row. Its final theorem accepts exactly stutterStep, with an exact structural gate delta and a fixed-machine affine emitted-gate bound. Canonical row validity, finite-label dispatch, and the complete local transition all expose such height-independent coefficients; the principal predicate builders also close to well-formed general circuits without changing evaluation. The fresh-layout layer allocates two consecutive nonaliasing rows at an arbitrary external-input offset, preserves a caller-supplied assignment outside them, and proves local completeness; its canonical wrapper exports the same finite assignment shape used by general-circuit satisfiability. Exact boundary constraints compare complete initial/accepting rows through one shared static pool, reject unencodable concrete targets by an actual false wire, and provide a symbolic-input-stack initial form. The normalized verifier sublayer packages an NP verifier with its concrete TM2 and polynomial certificate bound, then derives polynomial input, stuttering-horizon, and uniform stack-height envelopes without changing the underlying bounded run. Whole-tableau circuitization allocates every row, constrains row validity and adjacent transitions, recovers the bounded certificate shape, fixes exact initial and accepting boundaries, and closes their conjunction as a well-formed general circuit. Its satisfiability is equivalent to language membership, and its gate count has an explicit fixed-verifier polynomial bound. The validity layer builds canonical row validation with exact semantics and gate cost; its stack-validity index layer also gives closed affine wire formulas for every active-mask, cell, and ordered-stack output, and the circuitization layer now executes the five-gate XNOR subtrace at every fixed stack/cell coordinate together with the complete suffix-OR active mask of every fixed stack; the leading blank negation closes each semantic six-gate cell block. The workspace layer gives verified constant-cost widen/narrow bridges for one bundled transition.