§1.2, pp. 789–790 — the max-flow is upper bounded by the min-cut: f·|U||Ū| ≤ C(U, Ū) for a UMFP
ProvedLeightonRao.Uniform.weak_dualitymulticommodity-flowp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1sparsest-cutweak-duality
Let a network on nodes carry the uniform multicommodity flow problem (demand one for every unordered pair of nodes). If a concurrent flow of value exists, then for every nonempty proper subset ,
The commodities separated by the cut each send units across it, and all of that flow uses the capacity of the cut.
Dividing by gives for every cut, i.e. the max-flow is at most the min-cut — the upper half of Theorem 2.
Formalization Note Each unordered pair is two ordered commodities with demand (footnote 2, p. 791).
Preamble
import Definitions.Def_LeightonRao_Uniform_Flow set_option autoImplicit false open scoped BigOperators
Formal statement
namespace LeightonRao.Uniform
/-- The cut inequality of §1.2 specialized to the ordered-pair encoding of a UMFP. -/
theorem weak_duality {V : Type} [Fintype V] [DecidableEq V]
(N : Network V) (f : V → V → V → V → ℝ) (lam : ℝ)
(hf : IsConcurrentFlow N uniformDemand f lam)
(U : Finset V) (hU : U.Nonempty) (hUc : Uᶜ.Nonempty) :
lam * ((U.card : ℝ) * (Uᶜ.card : ℝ)) ≤ cutCap N U := by sorry
end LeightonRao.Uniform
Source
Leighton and Rao, Multicommodity max-flow min-cut theorems and their use in designing approximation algorithms, J. ACM 46 (1999), pp. 789–790, §1.2 (cut argument for f ≤ C(U, Ū)/D(U, Ū)), with D(U, Ū) = |U||Ū| from §1.5, p. 792
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.