Let G be an abelian group and let S⊆G be finite. For x∈G write
ΔS(x)=(x+S)∖S
for the number of elements that translating S by x moves out of S; equivalently ΔS(x)=∣S∣−∣(x+S)∩S∣, and in a finite ambient group ΔS(x)=∣(S+x)∩S∣. Then ΔS is subadditive:
ΔS(x+y)≤ΔS(x)+ΔS(y)for all x,y∈G.
The proof is the one-line containment
(x+y+S)∖S⊆((x+y+S)∖(y+S))∪((y+S)∖S),
together with the observation that (x+y+S)∖(y+S) is the translate by y of (x+S)∖S and therefore has the same cardinality.
This is the quantity Olson calls λ(g)=∣(B+g)∩B∣ in Section 4 of his 1975 paper, and its subadditivity is exactly what lets one transfer a large gain from an iterated sumset element d=c1+⋯+cn back to a single generator ci. It is the elementary half of Olson's Lemma 3.1 and of the corresponding step in the modern treatment of DeVos-Goddyn-Mohar-Šámal.
Formalization note. The translate x+S is written as S.image fun s => x + s, matching the notation used elsewhere in this mission for a+P(A). No finiteness of G is needed.
Preamble
import Definitions.Def_Erdos131_NonDividing
import Mathlib.Tactic
open Erdos131
open scoped Pointwise
Formal statement
theorem Erdos131.card_translate_sdiff_subadditive {G : Type*} [AddCommGroup G] [DecidableEq G]
(S : Finset G) (x y : G) :
((S.image fun s => (x + y) + s) \ S).card
≤ ((S.image fun s => x + s) \ S).card + ((S.image fun s => y + s) \ S).card := by sorry
Source
J. E. Olson, 'Sums of sets of group elements', Acta Arith. 28 (1975), 147-156, Section 4 (subadditivity of lambda(g) = |(B+g) cap complement(B)|, used in the proof of Lemma 3.1); the same statement appears as Observation 2.3, attributed to Erdos-Heilbronn, in M. DeVos, L. Goddyn, B. Mohar, R. Samal, 'A quadratic lower bound for subset sums', Acta Arith. 129 (2007), 187-195, arXiv:math/0612045, p. 6.