Endpoint and quantitative runner candidates

These modules remain unregistered. Five gates below have complete component receipts; the remaining gates are source candidates. Each uses the existing exact dependency-closure gate: source rebuild, an ownership-based audit of every project theorem, and independent replay with the pinned stock Lean kernel. Allowed axioms remain propext, Classical.choice, and Quot.sound.

The gates are separate so that an error in the finite enumeration does not prevent checking independent arithmetic or hyperplane arguments.

Task Statement to be checked
six-classification-bound The five-speed runner bound and primitive six-speed tight classification imply the six-speed runner bound for arbitrary positive lists of at most six entries.
hyperplane-endpoint-geometry For a primitive full-support normal, equality with the unshifted hyperplane maximum is equivalent to the actual velocity circle hitting the specified finite-order canonical residue.
hyperplane-endpoint-density The endpoint-equality event in the literal positive, pairwise-distinct box sample has limiting probability given by the totient and finite Euler product. Nonprimitive velocities remain in the sample.
hyperplane-rate-sharpness The normal (1,1,-1) and velocities g(1,2,3) give an actual value 1/4, a positive gap from the profile 1/3, and two-sided bounds of order 1/N for the uniform failure probability.
hyperplane-compact-subgroup Finite Fourier approximation and nontrivial-character cancellation control the actual integer velocity circles in a compact torus subgroup.
hyperplane-subspace-classification For a fixed shift and closed sublevel below the real-subspace maximum, the bad integer velocities form a finite union of proper integer-equation slices. The induction uses the torus closure of the actual subspace.
seven-quantitative-arithmetic Exact determinant, adjacent-ratio, and prefix inequalities give the area and rounding-error bounds at norm cutoff 7*42^10.
seven-rational-gap The real-time pair-sum maximum theorem gives a bounded denominator and an explicit gap above 1/7.
seven-minor-bound Actual integer minors and repeat-line generators are bounded using the first Gram--Schmidt residual and two-coordinate Cauchy--Schwarz.
seven-quantitative-basis A positive primitive seven-speed tuple above the cutoff lies in an actual two-dimensional rational prefix plane with the required area and value bounds.
seven-repeat-patterns Bounds on all actual repeated-speed directions imply the canonical signed six-speed patterns and normalized plane data.
seven-tangent-gap A failure of a complementary-pair relation produces an explicit point with clearance at least 229/1596.
seven-product-family Every primitive positive product list with base [1,2,3,4,5,6] and value below 1/7 has exact value s/(7*s+1) for an unrestricted integer s >= 1.
seven-product-directions Every positive integer direction in an actual signed standard product plane gives a permutation of the complete seven-entry product list.
seven-large-norm-family Every positive primitive seven-speed tuple with value below 1/7 and Euclidean norm at least 7*42^10 is a permutation of [1,2,3,4,5,6,7*s] and has value s/(7*s+1).

The geometric seven-speed statements retain LonelyRunnerBound 5 and PrimitiveSixClassification as explicit theorem inputs where required. They are not axioms, and these candidates do not complete the seven-speed result while those inputs remain undischarged. The small-norm alternatives also remain a separate obligation. The scalar product-family theorem has no lower-runner input; membership in that family is proved by other modules.

At 3f6ee74, the following complete closures passed source, axiom, and independent stock-kernel verification with unchanged sources:

Task Project modules Audited theorems
hyperplane-compact-subgroup 16 279
seven-quantitative-arithmetic 1 11
seven-rational-gap 5 94
seven-tangent-gap 12 203
seven-product-family 13 319

The exact jobs are in run 44468683790 and the machine-readable receipt. The minor bound and six-speed runner adapter subsequently passed repaired local source checks; those repairs still need their complete hosted gates.

The hyperplane subspace theorem is separate from exact quasipolynomial counting. No eventual counting statement is substituted for an all-positive- integer statement, and no general Ehrhart theorem is assumed.

Inspect a task without running Lean:

python3 scripts/verify_loneliness_final_extensions.py seven-large-norm-family --plan

Run the full source, axiom, and kernel gate for a task:

python3 scripts/verify_loneliness_final_extensions.py hyperplane-endpoint-density