Dense-pool intersection lemma: (Erdős 624 auxiliary)
OpenErdos624.dense_pool_intersectioncombinatoricserdos-problems
Let be a finite set with , and let be a finite colour set of size . Suppose each colour has a pool satisfying for some . For and , writing , the number of -subsets with satisfies .
Context: Erdős Problem 624 asks whether . This theorem does NOT settle Problem 624 / PAIR; it only proves Sol's dense-pool intersection counting lemma (6).
Formalization Note. Pure Mathlib: bOf a q = 2^(q+1)*a^q, poolInter/goodTuples as Finset filters; no dependence on a separate Erdős 624 definitions module.
Preamble
import Mathlib open Finset
Formal statement
namespace Erdos624
/-- `b = 2^{q+1} a^q`. -/
def bOf (a q : ℕ) : ℕ := 2 ^ (q + 1) * a ^ q
/-- Intersection of pools over `S` (empty → `univ`). -/
def poolInter {U C : Type*} [Fintype U] [DecidableEq U]
(V : C → Finset U) (S : Finset C) : Finset U :=
if h : S.Nonempty then S.inf' h V else univ
/-- Good q-tuples: those with `b · |∩ pools| ≥ n`. -/
def goodTuples {U C : Type*} [Fintype U] [DecidableEq U] [DecidableEq C]
(G : Finset C) (V : C → Finset U) (a q : ℕ) : Finset (Finset C) :=
let n := Fintype.card U
let b := bOf a q
{S ∈ G.powersetCard q | b * #(poolInter V S) ≥ n}
/-- Dense-pool intersection lemma (Sol (6)), multiplication-cleared.
Pools of density ≥1/a; for `g ≥ 2a(q−1)` and `n≥1`,
`b · q! · #good ≥ g^q` with `b = 2^{q+1} a^q`.
Does **not** settle PAIR / #624. -/
theorem dense_pool_intersection
{U C : Type*} [Fintype U] [DecidableEq U] [DecidableEq C]
(G : Finset C) (V : C → Finset U) (a q : ℕ)
(ha : 1 ≤ a) (hq : 1 ≤ q) (hn : 1 ≤ Fintype.card U)
(hPool : ∀ u ∈ G, a * #(V u) ≥ Fintype.card U)
(hg : 2 * a * (q - 1) ≤ #G) :
bOf a q * q.factorial * #(goodTuples G V a q) ≥ (#G) ^ q := by sorry
end Erdos624
Source
Erdős Problem 624 (https://www.erdosproblems.com/624); dense-pool intersection (Sol (6)) derived from GPT 6.1 Sol's analysis, relayed by Ethan Cheung; Lean-verified by Grok Bot for Ethan Cheung (Oct 24 2026).