Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Claim 2.2 — nearest centers and costs

Proved
TightCoreset2027.Euclidean.nearest_centers_and_costs

by mikedeng1 · Nov 9, 2026 · Mathlib 0df444a (Lean v4.33.1)

clusteringcomputational-geometrycoresetsp2o-batch-stoc27ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Take the §2.1 projection family, positive integer parameters n,m,q,Ln,m,q,Ln,m,q,L, the coding condition 2L>q+(q3)2^L>q+\binom q32L>q+(3q​), and n(n−1)+m(L+1)≤kn(n-1)+m(L+1)\le kn(n−1)+m(L+1)≤k. Let XXX be the nnn-copy dataset defined above. For every active copy bbb, direction v∈Vv\in Vv∈V and sign array σ\sigmaσ, the prescribed padded query CCC has exactly kkk distinct centers and ∣X∣=n2mq|X|=n^2mq∣X∣=n2mq. Every active point pb,j,i,sp_{b,j,i,s}pb,j,i,s​ at level ℓ=ℓ(s;v)\ell=\ell(s;v)ℓ=ℓ(s;v) has the translated ci,ℓc_{i,\ell}ci,ℓ​ as its unique nearest center, and

d(pb,j,i,s,C)2=Dℓ+n8L2σji(−1)ℓ,5n≤d(pb,j,i,s,C)2≤9n.d(p_{b,j,i,s},C)^2=D_\ell+\frac{\sqrt n}{8L^2}\sigma_{ji}(-1)^\ell,\qquad 5n\le d(p_{b,j,i,s},C)^2\le9n.d(pb,j,i,s​,C)2=Dℓ​+8L2n​​σji​(−1)ℓ,5n≤d(pb,j,i,s​,C)2≤9n.

Every inactive point has squared cost 222, and

cost⁡2(X,C)≤11∣X∣.\operatorname{cost}_2(X,C)\le11|X|.cost2​(X,C)≤11∣X∣.

The assertion verifies all assignment and cost clauses for the concrete centers, including every populated cell, without assuming their correctness in the setup.

Differences from the source:

  • encoding: sufficiently separated copies and sufficiently remote fixed padding → translations 20 sqrt(n) b f and padding 20 sqrt(n)(n+1+r) f; reason: the proof allows separation at least 20 sqrt(n); the fixed padding exceeds all copies and local query distances
  • encoding: orthogonal coordinate blocks and signs in {-1,1} → a Euclidean space indexed by the disjoint coordinate blocks and Boolean sign arrays; reason: the coordinate change is isometric and each sign array is represented
  • encoding: for every point in a populated cell, together with the constructed set sizes in the setup → indexed points with the specified level, and the set cardinalities included in the conclusion; reason: distinct triples parameterize the same cells; the size assertions are the explicit construction obligations from the preceding paragraphs
Preamble
import Definitions.Def_TightCoreset2027_Euclidean_queryConstruction
set_option autoImplicit false
Formal statement
namespace TightCoreset2027.Euclidean

theorem nearest_centers_and_costs (n m L q k : ℕ)
    (hn : 1 ≤ n) (hm : 1 ≤ m) (hL : 1 ≤ L) (hq : 1 ≤ q)
    (hcode : q + Nat.choose q 3 < 2 ^ L)
    (G : ProjectionFamily L q) (hG : ProjectionSpecification G)
    (hbudget : n * (n - 1) + m * (L + 1) ≤ k) :
    ∀ (b : Fin n) (v : ProjectionSpace L), v ∈ G.directions →
      ∀ σ : SignArray n m,
        let C := queryCenters n n m L k b (displacement2 L) v σ
        let X := dataset n n m G
        C.card = k ∧ X.card = n * n * m * q ∧
        (∀ (j : Fin n) (i : Fin m) (ℓ : Fin (L + 1)) (s : Fin q),
          G.level s v = ℓ →
            IsUniqueNearest (dataPoint G b.val j i s) C
              (activeCenter b.val (displacement2 L) v σ i ℓ) ∧
            nearestDistance (dataPoint G b.val j i s) C ^ 2 =
              baseline n L ℓ + Real.sqrt (n : ℝ) / (8 * (L : ℝ) ^ 2) *
                signValue (σ j i) * (-1 : ℝ) ^ ℓ.val ∧
            5 * (n : ℝ) ≤ nearestDistance (dataPoint G b.val j i s) C ^ 2 ∧
            nearestDistance (dataPoint G b.val j i s) C ^ 2 ≤ 9 * (n : ℝ)) ∧
        (∀ b' : Fin n, b' ≠ b → ∀ (j : Fin n) (i : Fin m) (s : Fin q),
          nearestDistance (dataPoint G b'.val j i s) C ^ 2 = 2) ∧
        cost 2 X C ≤ 11 * (X.card : ℝ) := by sorry

end TightCoreset2027.Euclidean
Source
Junqiu Hu, Lingxiao Huang, Jian Li, Baoduo Xu, STOC 2027 submission #879 (October 2026), Claim 2.2, Eq. (7)–(9), printed pp. 12–13 / PDF pp. 14–15; setup §§2.1–2.2, printed pp. 10–11 / PDF pp. 12–13; https://github.com/sigact-stoc/stoc2027/releases/download/papers-1/879.pdf
Human review
  • Endorsed by Shuze Chen · Nov 9, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Nov 9, 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 paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me