An asymptotic version of the prime power conjecture for perfect difference sets (Q2049954)

Let \(m \geq 1\). A set \(D \subset \mathbb{Z}/m\mathbb{Z}\) is called a perfect difference set if every non-zero \(a \in \mathbb{Z}/m\mathbb{Z}\) can be uniquely presented as a difference of two elements of \(D\). It is easy to see that if \(D\) is a perfect difference set, then necessarily \(m = n^2 + n + 1\) and \(|D| = n+1\) for some integer \(n\). For a perfect difference set \(D\), this integer \(n\) is called the order of \(D\). In addition to additive number theory, perfect difference sets are of interest in design theory and finite geometry. \textit{J. Singer} [Trans. Am. Math. Soc. 43, 377--385 (1938; Zbl 0019.00502; JFM 64.0972.04)] has constructed perfect difference sets whenever \(n\) is a prime-power, and prime power conjecture states that these are the only orders for which perfect difference sets exist. In the current paper, the author makes significant progress toward this conjecture by proving that the number of \(n \leq N\) such that \(n\) is an order of a perfect difference set is \((1+o(1))N/\log N\). The first step of the proof is to reduce the problem to studying the number of \(n\) for which certain relations hold between the prime factors of \(n\) and those of \(n^2+n+1\). The number of such \(n\) is then bounded using sieve methods and lattice point counting arguments.