Prove2Me
Navigate
DiscoverCollectionsFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Partitioner wins the rrr-partitioning game on every countable graph, for every ordinal rrr

Open
TenChallenges.challenge_10

by utkuokur · Nov 4, 2026 · Mathlib 0df444a (Lean v4.33.1)

infinite-graphsten-challengesunfriendly-partition

The scaled unfriendly partition conjecture. For every ordinal rrr and every graph GGG on a countable vertex set, Partitioner has a winning strategy in the rrr-partitioning game on GGG: starting from the empty partial partition with clock rrr, whenever Challenger names a vertex vvv and an ordinal β\betaβ below the current clock, Partitioner can extend the current finite partial partition to a finite partial partition that is unfriendly at vvv, and continue to do so with clock β\betaβ:

∀r∈Ord  ∀G countable:PartitionerWins(G,∅,r).\forall r\in\mathrm{Ord}\ \ \forall G \text{ countable}:\quad \mathrm{PartitionerWins}(G,\emptyset,r).∀r∈Ord  ∀G countable:PartitionerWins(G,∅,r).

An unfriendly partition of GGG yields such a strategy for every rrr, so the statement follows from the unfriendly partition conjecture for countable graphs. The statement is open for infinite clocks.

Formalization Note. The statement is ∀ r : Ordinal.{v}, statement_10.{u, v} r, with independent universes for the vertex type and the ordinal.

Preamble
import Definitions.Def_TenChallenges_Challenge_10
universe u v
Formal statement
namespace TenChallenges
theorem challenge_10 : ∀ r : Ordinal.{v}, statement_10.{u, v} r := by sorry
end TenChallenges
Source
Unfriendly partition conjecture (Cowan and Emerson), as stated in R. Aharoni, E. C. Milner, K. Prikry, Unfriendly partitions of a graph, JCTB 50 (1990), https://doi.org/10.1016/0095-8956(90)90092-E; game form in V. Bajpai, N. Bowler, J. Carmesin, U. Okur, 10 Scalable Challenges in Combinatorics, AI4Math @ CICM 2026, challenge 10
Read-back

What the Lean code literally says, in plain math · claude-fable-5-1

Setting and notation. Throughout, VVV is an arbitrary type (the vertex set) living in a universe uuu; VVV may be empty, finite or infinite. GGG is a simple graph on VVV, i.e. a binary adjacency relation v∼Gwv \sim_G wv∼G​w on VVV that is symmetric and irreflexive (no loops, no multiple edges, and no further restriction: a vertex may have infinitely many neighbours). NG(v)={ w∈V:v∼Gw }N_G(v) = \{\, w \in V : v \sim_G w \,\}NG​(v)={w∈V:v∼G​w} is the set of neighbours of vvv. A partial 2-colouring is any function f:V→{⊥,T,F}f : V \to \{\bot, \mathsf T, \mathsf F\}f:V→{⊥,T,F}, where ⊥\bot⊥ means "no colour assigned" and T,F\mathsf T, \mathsf FT,F are the two Boolean colours; for a Boolean bbb, bˉ\bar bbˉ denotes its negation (Tˉ=F\bar{\mathsf T} = \mathsf FTˉ=F, Fˉ=T\bar{\mathsf F} = \mathsf TFˉ=T). ∣X∣|X|∣X∣ denotes the cardinality of XXX as a cardinal number (in universe uuu), so it may be infinite, and ≤\le≤ between cardinalities means existence of an injection. Ordinals α,β,r\alpha, \beta, rα,β,r range over the ordinals of a second universe vvv, independent of uuu; β<α\beta < \alphaβ<α is the usual strict well-ordering of ordinals.

Theorem challenge_10. The theorem asserts

∀r ordinal (universe v):S10(r),\forall r \ \text{ordinal (universe } v):\quad S_{10}(r),∀r ordinal (universe v):S10​(r),

that is, fully unfolded:

∀r,  ∀V (a type in universe u),  V countable  ⟹  ∀G simple graph on V:  W(G, f∅, r).\forall r,\ \ \forall V\ (\text{a type in universe } u),\ \ V \text{ countable} \;\Longrightarrow\; \forall G \text{ simple graph on } V:\ \ W\bigl(G,\, f_\emptyset,\, r\bigr).∀r,  ∀V (a type in universe u),  V countable⟹∀G simple graph on V:  W(G,f∅​,r).

Countable means there exists an injective map V→NV \to \mathbb NV→N (so VVV may be countably infinite, finite, or empty). The simple graph GGG is arbitrary subject only to being symmetric and irreflexive (vertices may have infinitely many neighbours; nothing else is assumed). Here f∅f_\emptysetf∅​ is the all-uncoloured partial colouring, f∅(v)=⊥f_\emptyset(v) = \botf∅​(v)=⊥ for every vvv, and W(G,f,α)W(G, f, \alpha)W(G,f,α) ("Partitioner wins from (f,α)(f,\alpha)(f,α)") is the proposition defined by well-founded recursion on the ordinal α\alphaα by

W(G,f,α)  ⟺  ∀v∈V,  ∀β<α,  ∃ g:V→{⊥,T,F}:L(G,f,g,v) ∧ W(G,g,β),W(G,f,\alpha) \iff \forall v \in V,\ \ \forall \beta < \alpha,\ \ \exists\, g : V \to \{\bot,\mathsf T,\mathsf F\}:\quad L(G,f,g,v)\ \wedge\ W(G,g,\beta),W(G,f,α)⟺∀v∈V,  ∀β<α,  ∃g:V→{⊥,T,F}:L(G,f,g,v) ∧ W(G,g,β),

where the legal-response predicate LLL is the conjunction

L(G,f,g,v)  ⟺  (∀w∈V, ∀b∈{T,F}: f(w)=b⇒g(w)=b)⏟g extends f ∧ { w∈V:g(w)≠⊥ } is finite⏟g has finite domain ∧ Unf(G,g,v)⏟g is unfriendly at v,L(G,f,g,v) \iff \underbrace{\bigl(\forall w \in V,\ \forall b \in \{\mathsf T,\mathsf F\}:\ f(w) = b \Rightarrow g(w) = b\bigr)}_{g \text{ extends } f} \ \wedge\ \underbrace{\{\, w \in V : g(w) \ne \bot \,\} \text{ is finite}}_{g \text{ has finite domain}} \ \wedge\ \underbrace{\mathrm{Unf}(G,g,v)}_{g \text{ is unfriendly at } v},L(G,f,g,v)⟺g extends f(∀w∈V, ∀b∈{T,F}: f(w)=b⇒g(w)=b)​​ ∧ g has finite domain{w∈V:g(w)=⊥} is finite​​ ∧ g is unfriendly at vUnf(G,g,v)​​,

and unfriendly at vvv means

Unf(G,g,v)  ⟺  ∃ b∈{T,F}: g(v)=b  and  ∣{ w:v∼Gw, g(w)=b }∣  ≤  ∣{ w:v∼Gw, g(w)=bˉ }∣,\mathrm{Unf}(G,g,v) \iff \exists\, b \in \{\mathsf T,\mathsf F\}:\ g(v) = b\ \text{ and }\ \bigl|\{\, w : v \sim_G w,\ g(w) = b \,\}\bigr| \;\le\; \bigl|\{\, w : v \sim_G w,\ g(w) = \bar b \,\}\bigr|,Unf(G,g,v)⟺∃b∈{T,F}: g(v)=b  and  ​{w:v∼G​w, g(w)=b}​≤​{w:v∼G​w, g(w)=bˉ}​,

i.e. vvv is coloured by ggg and, comparing cardinal numbers (possibly infinite), the neighbours of vvv with the same colour as vvv are no more numerous than the neighbours with the opposite colour; uncoloured neighbours count on neither side. Spelled out, W(G,f,α)W(G,f,\alpha)W(G,f,α) says: for every vertex vvv whatsoever (coloured or not; the same vertex may be chosen repeatedly) and every ordinal β\betaβ strictly below α\alphaα (all of them, not only a predecessor), there is a partial colouring ggg, which may depend on both vvv and β\betaβ, that keeps every colour already assigned by fff, colours only finitely many vertices in total, is unfriendly at vvv (nothing is required at other vertices, and ggg need not differ from fff), and from which the same property holds again with clock β\betaβ. Since the clocks strictly decrease, every such sequence of challenges is finite, and W(G,g,0)W(G,g,0)W(G,g,0) holds unconditionally (there is no β<0\beta < 0β<0).

The statement is polymorphic in the two universe levels uuu (where the vertex types live) and vvv (where the ordinals live): it is a family of assertions, one for every choice of these levels, and the quantifier over rrr ranges over all ordinals of universe vvv, with no bound. Edge cases literally included: for r=0r = 0r=0 the assertion is trivially true; for empty VVV it is trivially true for every rrr; for every r>0r > 0r>0 and every nonempty countable VVV and every simple graph on it, it demands a finite-domain colouring unfriendly at any challenged vertex, and then recursively for every smaller clock. The declaration's proof is left as a placeholder (it is stated, not proved); this read-back concerns only the statement.

Human review
  • Endorsed by Shuze Chen · Nov 4, 2026

    Confirmed by the moderator at approval.

  • Endorsed by utkuokur · Nov 4, 2026

    Confirmed by the mission captain (proposal self-audit).

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