Avoiding arithmetic progressions in cyclic groups (Q817414)

Summary: The function \(\alpha(n, r)\) is monotone in \(r\) but not monotone in \(n\). However, for any positive integers \(n\), \(m\) and \(r\) we have \(\alpha(nm, r)\geq\alpha(n, r)\cdot\alpha(m, r)\). In particular, for any positive integers \(n\), \(k\), and \(r\) we have \(\alpha(n^k, r)\geq\alpha(n, r)^k\), which implies that for any prime number \(p\), \(\alpha(p^k, p)\geq\alpha(p, p)^k= (p-1)^k\). On the one hand, for each prime number \(p> 2\) there are integers \(k\) such that \(\alpha(p^k, p)> (p- 1)^k\), but on the other hand, for every prime number \(p\) we have \(\alpha(p^2, p)= (p-1)^2\) and \(\alpha(p, p)= (p- 1)\). In addition, we have seen that \(\alpha(3^3, 3)= 2^3\) (Proposition 2.7) but the authors were not able to prove \(\alpha(5^3, 5)= 4^3\), since the proof of Proposition 2.7 seems not generalisable. This leads to the open question whether \(\alpha(p^3, p)= (p- 1)^3\) for all primes \(p\) larger than \(3\) (the authors could not agree what they expect to be the answer). Further, we have seen that for any positive integer \(n\), \(\alpha(n, n)= n(1-{1\over p})\), where \(p\) is the smallest prime number dividing \(n\). In particular, for any positive integer \(m\) we have \(\alpha(2m, 2m)= m\).