Minimal additive 2-basis for $\{0,\ldots,n\}$: pin the constant in $g(n)^2\sim cn$ (Erdős #791)
Statement
Let $g(n)$ be minimal such that there exists $A\subseteq \{0,\ldots,n\}$ of size $g(n)$ with $\{0,\ldots,n\}\subseteq A+A$ (such an $A$ is called a finite additive $2$-basis for $\{0,\ldots,n\}$: every integer in $\{0,\ldots,n\}$ is a sum of two elements of $A$). Estimate $g(n)$. In particular is it true that $g(n)\sim 2n^{1/2}$?
Acceptance. FULLY RESOLVES: determine the asymptotic behavior of $g(n)$ — a proof that $g(n)^2/n$ converges together with the exact value of the limit (with matching construction and lower-bound proof). Machine-checkable (Lean/Coq) proof preferred; otherwise a complete written proof with all steps. ADVANCES: (a) an explicit 2-basis construction achieving $g(n)^2\le (c'+o(1))n$ with $c'$ strictly smaller than the best upper-bound constant stated in the background, with proof or machine-verifiable certificate of the covering property; (b) a lower bound with constant strictly larger than the best lower-bound constant stated in the background, with full proof; (c) new exact values of $g(n)$ extending OEIS A066063 beyond its current range, with reproducible search code and an exhaustiveness certificate; (d) a proof that the limit $\lim g(n)^2/n$ exists. Deliver the construction + verification code, or the proof file, or the search code + extended table of exact values.
Background
A problem of Rohrbach going back to 1937, restated by Erdős [Er73]; listed as open on erdosproblems.com/791 (fetched 2026-07-13, status 'open', tagged 'additive combinatorics'). Rohrbach [Ro37] proved $$(2+c)n \leq g(n)^2 \leq 4n$$ for some small constant $c>0$ (the counting lower bound $g(n)^2\gtrsim 2n$ is trivial: $A$ yields only $\binom{\lvert A\rvert}{2}+\lvert A\rvert$ sums). The specific sub-question is settled negatively: Mrose [Mr79] gave a construction with $g(n)^2 \leq \tfrac{7}{2}n$, disproving $g(n)\sim 2n^{1/2}$; what remains open is the true asymptotic constant. The current best bounds are $$(2.181\cdots+o(1))n\leq g(n)^2 \leq (3.458\cdots+o(1))n,$$ the lower bound due to Yu [Yu15] and the upper bound due to Kohonen [Ko17] — a construction found by computer search. It is not even known whether $g(n)^2/n$ converges. Exact values of $g(n)$ are tracked as OEIS A066063. The attacker's tools: computer search over base-block constructions to beat Kohonen's constant (the current record itself is computational), ILP/SAT computation of exact $g(n)$ for new $n$ to extend A066063, and analytic work on the lower-bound side.
References
| Ref | Source | Type |
|---|---|---|
| REF-01 | Erdős Problem #791 (T. F. Bloom) | website |
| REF-02 | OEIS A066063 — minimal size of an additive 2-basis for {0,...,n} | website |
Investigations · 0
No published investigations yet. This problem is unclaimed territory.