Bounded polyhedral images admit uniform bounded nonnegative representatives
ProvedHirsch.bounded_image_compact_representativesFor every finite real resource map B and linear image map G, there is a positive constant C depending only on B and G such that every nonnegative x with Bx<=t has a nonnegative representative y with By<=t and Gy=Gx, total mass at most C*(norm(t)+norm(Gx)), and every coordinate-zero or tight resource row of x remains so at y. Consequently every uniformly bounded image of a resource polyhedron is unchanged after adding one sufficiently large total-mass bound; that capped coefficient set is compact. Neither original coefficient boundedness, injectivity of G, a coverage certificate, recession annihilation, nonemptiness, a vertex decomposition nor a right-inverse oracle is assumed. Constants are not claimed bounded by dimensions or rational bit complexity. This gives a compact representation of the same shape, not an ordinary-edge or diameter transport theorem.
import Mathlib open Set open scoped BigOperators
namespace Hirsch
theorem bounded_image_compact_representatives {k r d : ℕ}
(B : (Fin k → ℝ) →L[ℝ] (Fin r → ℝ))
(G : (Fin k → ℝ) →L[ℝ] (Fin d → ℝ)) :
(∃ C : ℝ, 0 < C ∧ ∀ (t : Fin r → ℝ) (x : Fin k → ℝ),
(∀ i, 0 ≤ x i) → (∀ q, B x q ≤ t q) →
∃ y : Fin k → ℝ, (∀ i, 0 ≤ y i) ∧ (∀ q, B y q ≤ t q) ∧
G y = G x ∧ (∀ i, x i = 0 → y i = 0) ∧
(∀ q, B x q = t q → B y q = t q) ∧
(∑ i, y i) ≤ C * (‖t‖ + ‖G x‖)) ∧
(∀ (t : Fin r → ℝ) (M : ℝ), 0 ≤ M →
(∀ x : Fin k → ℝ, (∀ i, 0 ≤ x i) →
(∀ q, B x q ≤ t q) → ‖G x‖ ≤ M) →
∃ R : ℝ, 0 ≤ R ∧
IsCompact {x : Fin k → ℝ | (∀ i, 0 ≤ x i) ∧
(∀ q, B x q ≤ t q) ∧ (∑ i, x i) ≤ R} ∧
G '' {x : Fin k → ℝ | (∀ i, 0 ≤ x i) ∧ ∀ q, B x q ≤ t q} =
G '' {x : Fin k → ℝ | (∀ i, 0 ≤ x i) ∧
(∀ q, B x q ≤ t q) ∧ (∑ i, x i) ≤ R}) := by sorry
end Hirsch