Answers are grounded in this paper and cite their sources.
Theoretical analysis establishes upper bounds for separated B3 and B4 real sumsets, matching the best known integer constants without asymptotic penalty.
Call a finite set S of real numbers a 1-separated B_h set if any two distinct h-element multisets from S have sums differing by at least 1. These are the real, unit-separation cases of the Bh,Δ-sets recently introduced by Nathanson. We prove that a 1-separated B₃ set S ⊆ [0,V] satisfies |S| ≤ (2/μ₂²)1/3 V1/3 (1+o(1)) < 1.51547 V1/3 as V → ∞, and that a 1-separated B₄ set satisfies |S| ≤ (4/μ₂²)1/4 V1/4 (1+o(1)) < 1.62431 V1/4, where μ₂² is the L² autoconvolution constant. These are the first upper bounds in the h=3 and h=4 cases of Nathanson's extension problem for Bh,Δ-sets to improve on trivial packing-type counting, and the constants coincide with the best known bounds for ordinary integer B₃ and B₄ sets, due to White: for these two cases, at the current frontier of knowledge, the separated-real relaxation costs nothing asymptotically. The proof combines windowed counting lemmas, in which the separation hypothesis reproduces exactly the multiset rigidity available in the integer case, with a Fejér-kernel smoothing argument that feeds the autoconvolution constant directly, bypassing the variational machinery of Green.
No takes yet. Share an insight, caveat, or question.
Ryu Ogawa (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: