Submodularity of the cut function
ProvedLysgaardCVRP.Shrink.cut_submodulargraph-cutsp2o-batch-p200bp2o-gran-per-chapterp2o-plan-paperp2o-v1submodularity
Let be a nonnegative edge vector on the complete graph with vertex set . For any two vertex sets ,
Equivalently, : the cut function of a nonnegatively weighted graph is submodular. The paper quotes this classical fact (Nemhauser and Wolsey, 1988, p. 660) in the proof of Proposition 1, in the arrangement displayed above.
Formalization Note Stated for arbitrary vertex sets, including sets containing the depot. Nonnegativity of is necessary.
Preamble
import Mathlib import Definitions.Def_LysgaardCVRP_Shrink_cut
Formal statement
namespace LysgaardCVRP.Shrink
/-- Submodularity of the cut function, in the arrangement used in the proof of Proposition 1 of
Lysgaard, Letchford & Eglese, *A new branch-and-cut algorithm for the capacitated vehicle
routing problem*, Math. Program. Ser. A 100 (2004), p. 426 (PDF p. 4) (unnumbered; the paper cites Nemhauser & Wolsey, *Integer and
Combinatorial Optimization*, 1988, p. 660): "It follows from the submodularity of the cut function
that $x^*(\delta(T)) - x^*(\delta(S\cup T)) \ge x^*(\delta(S\cap T)) - x^*(\delta(S))$".
**Formalization Note.** Stated for arbitrary vertex sets `S`, `T` of the complete graph on
`Fin (n+1)` (the fact does not need the depot condition; the paper applies it to customer sets),
and for a nonnegative edge vector $x \ge 0$ (the LP bounds of (3)–(4)); it is false for signed
$x$. -/
theorem cut_submodular {n : ℕ} (x : Sym2 (Fin (n + 1)) → ℝ) (hx : ∀ e, 0 ≤ x e)
(S T : Finset (Fin (n + 1))) :
cut x T - cut x (S ∪ T) ≥ cut x (S ∩ T) - cut x S := by sorry
end LysgaardCVRP.Shrink
Source
Lysgaard, Letchford & Eglese, A new branch-and-cut algorithm for the capacitated vehicle routing problem, Math. Program. Ser. A 100 (2004), p. 426 (PDF p. 4), proof of Proposition 1 (citing Nemhauser & Wolsey 1988, p. 660)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.