Polynomial edge refinement of circuit walks with a high-dimensional common face
OpenHirsch.polynomial_edge_refinement_of_circuit_walks_of_high_carrierPolynomial edge refinement of circuit walks that contain at least one high-dimensional common face.
Let and let be a bounded H-polytope given by an irredundant, strictly feasible family of inequalities. Suppose and are vertices joined by a padded row-circuit walk for which some consecutive pair has common-face dimension
The claim is that there exist constants , independent of and of the walk, such that and are also joined by a padded vertex-edge walk of length .
This is the complementary case to routing through common faces of dimension at most three, which is already reduced to Klee's theorem. The present statement isolates the remaining graph-routing difficulty: a circuit displacement whose common carrier has dimension four or more. Constant or dimension-only overhead per circuit step is already false for polygons; polynomial row-dependent overhead is not ruled out.
Formalization Note The extra hypothesis is existential over the given walk, not a restriction of the ambient polytope. Walks are padded.
import Definitions.Def_Hirsch_circuit_model import Definitions.Def_Hirsch_common_face_geometry set_option autoImplicit false open scoped RealInnerProductSpace open Hirsch
theorem Hirsch.polynomial_edge_refinement_of_circuit_walks_of_high_carrier :
∃ C k : ℕ, ∀ (d n : ℕ)
(a : Fin n → EuclideanSpace ℝ (Fin d)) (b : Fin n → ℝ),
4 ≤ d →
Bornology.IsBounded (Hirsch.Hpoly a b) →
Hirsch.RowPresentationIrredundant a b → Hirsch.StrictlyFeasibleRows a b →
∀ u ∈ Set.extremePoints ℝ (Hirsch.Hpoly a b),
∀ v ∈ Set.extremePoints ℝ (Hirsch.Hpoly a b),
∀ L : ℕ, ∀ w : ℕ → EuclideanSpace ℝ (Fin d),
w 0 = u → w L = v →
(∀ j ≤ L, w j ∈ Hirsch.Hpoly a b) →
(∀ j < L, w j = w (j + 1) ∨ Hirsch.RowCircuitStep a b (w j) (w (j + 1))) →
(∃ j < L, 3 < HirschCommonFace.commonFaceDim a b (w j) (w (j + 1))) →
∃ q : ℕ → EuclideanSpace ℝ (Fin d),
q 0 = u ∧ q (C * (n + d) ^ k * L) = v ∧
∀ j < C * (n + d) ^ k * L,
q j = q (j + 1) ∨
Hirsch.Adj (Hirsch.Hpoly a b) (q j) (q (j + 1)) := by sorry