Erdős Problem 402 #
Reference: erdosproblems.com/402
A conjecture of Graham [Gr70], who also conjectured that (assuming $A$ itself has no common divisor) the only cases where equality is achieved are when $A = \{1, ..., n\}$ or $\{L/1, ..., L/n\}$ (where $L = \operatorname{lcm}(1, ..., n)$) or $\{2, 3, 4, 6\}$. Note: The source [BaSo96] mentioned on the Erdős page makes it clear what quantfiers to use for "where equality is achieved". See Theorem 1.1 there.
TODO(firsching): Consider if we should have the other direction here as well or an iff statement.
Proved for all sufficiently large sets (including the sharper version which characterises the case of equality) independently by Szegedy [Sz86] and Zaharescu [Za87]. The following is taken from [Sz86].
There exists an effectively computable $n_0$ with the following properties: (i) if $n\geq n_0$ and $a_1, a_2, ..., a_n$ are distinct natural numbers then $\max_{i, j} \frac{a_i}{(a_i, a_j)} \geq n$. (ii) If equality holds then the system $\{a_1, a_2, ..., a_n\}$ is either of the type $\{k, 2k, ..., nk\}$ or of the type $\left\{\frac{k}{1}, \frac{k}{2}, ..., \frac{k}{n}\right\}$