Maximal -adic valuation of a unit-fraction representation of is attained twice
ProvedErdos287.padic_multiplicityLet and let be integers, each , with
Then for every prime and every index there is a different index with , where denotes the -adic valuation.
Equivalently: for each prime , the largest power of that divides one of the denominators divides at least two of them. The proof is the ultrametric one-term-dominates argument. If a single denominator were strictly the most divisible by , then its term would have -adic norm strictly larger than that of every other term, so the sum would satisfy . For this contradicts , and for the hypothesis is already vacuous because valuations are non-negative.
Neither monotonicity nor distinctness of the denominators is needed. This is the engine behind the classical theorem of Kürschák and Erdős on blocks of consecutive integers, and it is the basic structural constraint available for Erdős problem #287.
import Mathlib
namespace Erdos287
theorem padic_multiplicity (k : ℕ) (hk : 2 ≤ k) (f : ℕ → ℕ)
(hf1 : ∀ i, i < k → 1 < f i)
(hsum : ∑ i ∈ Finset.range k, (1 : ℚ) / f i = 1)
(p : ℕ) (hp : Nat.Prime p) (i : ℕ) (hi : i < k) :
∃ j, j < k ∧ j ≠ i ∧ padicValNat p (f i) ≤ padicValNat p (f j) := by sorry
end Erdos287