Prove2Me
Navigate
DiscoverFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Splitting off the Steiner vertices of an even multigraph

Disproved
MetricTSP.eliminate_steiner

by Shuze Chen · Aug 24, 2026 · Mathlib c5ea003 (Lean v4.30.0)

combinatorial-optimizationconnectivitygraph-theory

Lovász splitting-off, metric form. Let HHH be a multigraph (a symmetric multiplicity function with zero diagonal) on the cities with all degrees even, let TTT be a set of terminals, and suppose every cut separating two terminals has at least 2N2N2N edges. Then all the non-terminal (Steiner) vertices can be split off: there is a multigraph H′H'H′ whose edges lie inside TTT, with the same degrees on TTT, still with every terminal-separating cut of size at least 2N2N2N, and, since splitting a path u−v−wu{-}v{-}wu−v−w into the edge u−wu{-}wu−w does not increase metric cost, with

∑u,vH′(u,v) c(u,v)  ≤  ∑u,vH(u,v) c(u,v).\sum_{u,v} H'(u,v)\,c(u,v) \;\le\; \sum_{u,v} H(u,v)\,c(u,v).u,v∑​H′(u,v)c(u,v)≤u,v∑​H(u,v)c(u,v).

The combinatorial core is Lovász's splitting-off lemma in its Eulerian form: at a vertex vvv of even degree, some pair of edges at vvv can be replaced by their shortcut while preserving all local edge-connectivities among the other vertices; the uniform-demand version needed here (preserve all terminal-separating cuts at level 2N2N2N) follows from submodularity of the cut function by the standard analysis of dangerous sets. Iterating at one Steiner vertex reduces its degree to zero, and induction removes them all.

Combined with MetricTSP.even_set_matching on the resulting terminal-supported multigraph, this supplies the parity-correction matching (MetricTSP.parity_matching) of Wolsey's 32\tfrac3223​ analysis.

Preamble
import Mathlib
import Definitions.Def_MetricTSP_model
Formal statement
namespace MetricTSP

theorem eliminate_steiner (n : ℕ) (c : Fin n → Fin n → ℝ) (hc : IsMetricCost c)
    (T : Finset (Fin n)) (N : ℕ) (hN : 1 ≤ N)
    (H : Fin n → Fin n → ℕ) (hsym : ∀ u v, H u v = H v u) (hdiag : ∀ v, H v v = 0)
    (heven : ∀ v, Even (∑ u, H v u))
    (hcut : ∀ S : Finset (Fin n), (S ∩ T).Nonempty → (Sᶜ ∩ T).Nonempty →
      2 * N ≤ ∑ u ∈ S, ∑ v ∈ Sᶜ, H u v) :
    ∃ H' : Fin n → Fin n → ℕ, (∀ u v, H' u v = H' v u) ∧ (∀ v, H' v v = 0) ∧
      (∀ u v, H' u v ≠ 0 → u ∈ T ∧ v ∈ T) ∧
      (∀ v ∈ T, ∑ u, H' v u = ∑ u, H v u) ∧
      (∀ S : Finset (Fin n), (S ∩ T).Nonempty → (Sᶜ ∩ T).Nonempty →
        2 * N ≤ ∑ u ∈ S, ∑ v ∈ Sᶜ, H' u v) ∧
      ∑ u, ∑ v, (H' u v : ℝ) * c u v ≤ ∑ u, ∑ v, (H u v : ℝ) * c u v := by sorry

end MetricTSP
Source
L. Lovász, On some connectivity properties of Eulerian graphs, Acta Mathematica Academiae Scientiarum Hungaricae 28 (1976) 129-138, https://doi.org/10.1007/BF01902503 (the Eulerian splitting-off lemma); A. Frank, Connections in Combinatorial Optimization, Oxford University Press 2011, Chapter 8 (splitting-off techniques).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paper
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me