Olson: a zero-sum-free set has more than subset sums
ProvedErdos131.olson_card_subsetSumsLet be an abelian group and let be a finite, nonempty set of distinct elements which is zero-sum-free: no nonempty subset of sums to . Write
for the set of all subset sums of ; the empty subset contributes the element , so always. Then
This is a theorem of J. E. Olson, quoted as Theorem 7 by Erdős, Lev, Rauzy, Sándor and Sárközy. It is the engine behind every bound of the shape "a zero-sum-free set in a finite group is small": since is contained in , the inequality immediately gives for a zero-sum-free subset of a finite abelian group . It is one of the two classical addition theorems on which the Erdős–Lev–Rauzy–Sándor–Sárközy bound for non-dividing sets rests.
Formalization Note The set of subset sums is rendered as the image of the powerset of under the summation map, and the zero-sum-free hypothesis quantifies over members of A.powerset in the same style as the definition of NonDividing. Nonemptiness of is required: for both sides of the inequality equal , so the strict inequality fails. No finiteness assumption on is imposed, matching the source, which states the result for an arbitrary abelian group.
import Definitions.Def_Erdos131_NonDividing import Mathlib.Tactic open Erdos131
theorem Erdos131.olson_card_subsetSums {G : Type*} [AddCommGroup G] [DecidableEq G]
(A : Finset G) (hA : A.Nonempty)
(hzs : ∀ S ∈ A.powerset, S.Nonempty → (∑ x ∈ S, x) ≠ 0) :
1 + (A.card : ℝ) ^ 2 / 9 < ((A.powerset.image fun S => ∑ x ∈ S, x).card : ℝ) := by sorry