An arithmetic theorem related to groups of bounded nilpotency class. (Q2496845)

Let us define a multiplicative function \(\psi\colon\mathbb{N}\to\mathbb{N}\) via \(\psi(1)=1\) and \(\psi(p^\nu)=(p^\nu-1)(p^{\nu-1}-1)\cdots(p-1)\) (\(p\) prime, \(\nu>0\)). The main result is the following Theorem 1. Let \(m,c\in\mathbb{N}\). Then the following assertions are equivalent: (a) \((m,\psi(m))=1\) and \(m\) is \((c+2)\)-free, (b) Every group of order \(m\) is nilpotent of class at most \(c\). Reviewer's remark: The proof is too long. Indeed, the hypothesis is inherited by subgroups, so, if \(G\) is a minimal counterexample, it is minimal nonnilpotent of order, say \(p^aq^b\) with \(|G'|=q^b\). By Sylow, \(p\mid\psi(q^b\mid\psi(m))\), a contradiction. Thus, \(G\) is nilpotent. Next, \(m\) must be \((c+2)\)-free since, for every \(p\), there exists a \(p\)-group of order \(p^{c+2}\) and class \(c+1\). The converse may be extracted from the above argument.