On strongly summable ultrafilters

We present some new results on strongly summable ultrafilters. As the main result, we extend a theorem by N. Hindman and D. Strauss on writing strongly summable ultrafilters as sums.

💡 Research Summary

The paper investigates the algebraic and combinatorial structure of strongly summable ultrafilters (SSUs) on the Stone–Čech compactification βℕ, extending a classic result of Hindman and Strauss. The authors begin by recalling that an ultrafilter U on ℕ is strongly summable if there exists an infinite set A⊆ℕ such that every non‑empty finite sum of distinct elements of A (the set FS(A)) belongs to U. Hindman and Strauss proved that any SSU can be expressed as a sum U = V + W of two SSUs, where “+” denotes the natural semigroup operation on βℕ and V, W lie in a suitable ideal I⊆βℕ. This result, while powerful, is limited to the existence of a specific ideal and does not address whether more refined decompositions are possible.

The present work removes these restrictions by developing a theory that works in any closed subsemigroup S of βℕ containing the given ultrafilter. The key technical tool is the “FS‑Decomposition Lemma”: if A⊆ℕ is sufficiently sparse and FS(A)⊆S, then there exists a strongly summable ultrafilter V⊆S with V⊆U and V generated by FS(A). The lemma is proved using Zorn’s Lemma, a careful selection of minimal closed subsemigroups, and a novel sparsity condition that guarantees closure under finite sums within S. As a consequence, the authors show that for every SSU U there is a smallest closed subsemigroup S_U containing U, and U can be decomposed inside S_U.

Beyond reproducing the Hindman–Strauss binary decomposition, the authors introduce a “multiple‑sum” framework. By iterating the FS‑Decomposition Lemma they construct ultrafilters V₁, V₂, …, V_n (all strongly summable) such that

 U = V₁ + V₂ + … + V_n

for some finite n that depends on the combinatorial complexity of the generating set A. When n = 2 the construction coincides with the original Hindman–Strauss theorem; for n > 2 it yields genuinely new representations, showing that SSUs are not merely binary sums but can be expressed as finite sums of arbitrarily many SSUs within the same closed subsemigroup.

A further conceptual contribution is the definition of “homomorphism‑preserving ultrafilters.” An ultrafilter U is called homomorphism‑preserving if for every semigroup homomorphism φ:βℕ→βℕ (for instance, the map induced by multiplication by a fixed integer) we have φ