A criterion for HNN extensions of finite \(p\)-groups to be residually \(p\). (Q538081)

In this paper the authors deal with HNN-extensions of finite \(p\)-groups and give a criterion to be residually \(p\). Before stating the main result of the paper we quote some terminology. Let \(G^*=\langle G,t\mid t^{-1}At=B\varphi\rangle\) be the HNN-extension of the group \(G\) with associated subgroups \(A\) and \(B\) via the isomorphism \(\varphi\colon A\to B\). Let \(H_0=A\cap B\) and inductively \(H_{i+1}=\varphi^{-1}(H_i)\cap H_i\cap\varphi(H_i)\). If \(H(G,\varphi)=\bigcap_iH_i\), then \(\varphi\) restricts to an automorphism of \(H(G,\varphi)\) and this is the largest subgroup of \(G\) with this property. For the HNN-extension \(G^*\) a central compatible filtration is a finite sequence \(G=G_1\triangleright G_2\triangleright\cdots\triangleright G_n=1\) of normal subgroups of \(G\) such that \(G_i/G_{i+1}\) is central in \(G/G_{i+1}\) for each \(i\) and \(\varphi\) restricts to an isomorphism \(A\cap G_i\to B\cap G_{i+1}\), for each \(i\). Proposition: Let \(G^*=\langle G,t\mid t^{-1}At=B\varphi\rangle\) be the HNN-extension of the finite \(p\)-group \(G\). If \(G^*\) is residually \(p\), then there exists a central compatible filtration \(\mathbb G=(G_i)\) of \(G\), such that for any \(i<j\), any \(a\in A\cap G_i\) and any \(b\in B\cap G_j\) the automorphism of \(H(G_i/G_j,\tau_b\circ\varphi_{ij}\circ\tau_a)\) induced by \(\tau_b\circ\varphi_{ij}\circ\tau_a\) is a power of \(p\). (\(\tau_a,\tau_b\) are the inner automorphisms induced by \(a\) and \(b\), respectively). Theorem: Let \(G^*\) as above. Suppose there exists a central compatible filtration \(\mathbb G=(G_i)\) of \(G\), such that for any \(i\) the order of the automorphism of \(H(G_i/G_{i+1},\varphi_{i,i+1})\) induced by \(\varphi_{i,i+1}\) is a power of \(p\). Then \(G^*\) is residually \(p\). The paper, as the authors admit, is inspired by the ideas of \textit{E. Raptis} and \textit{D. Varsos} [in Pure Appl. Algebra 76, No. 2, 167-178 (1991; Zbl 0743.20022)]. But in that paper there is an oversight in an inductive step. Here the authors point it out and reconstruct the one direction of the Theorem 13 there as in the Theorem above.