Signalizer lattices in finite groups. (Q1024871)

The following question was suggested by a theorem of Pálfy and Pudlák: is each nonempty finite lattice isomorphic to an overgroup lattice \(\mathcal O_G(H)\) for some finite group \(G\) and subgroup \(H\) of \(G\)? The answer is almost certainly negative. However the question has remained open for almost 30 years since the paper of Pálfy and Pudlák was published. In a previous paper the author started to deal with this question, showing that the structure of \(G\) is highly restricted if the lattice \(\mathcal O_G(H)\) is ``disconnected''. He defined the notion of a ``signalizer lattice'' and showed that for suitable disconnected lattices \(\Lambda\) (the CD-lattices), if \(G\) is minimal subject to \(\mathcal O_G(H)\) being isomorphic to \(\Lambda\) or its dual, then either \(G\) is almost simple or \(H\) admits a signalizer lattice isomorphic to \(\Lambda\) or its dual. In this paper the author initiates the study of signalizer lattices, with the hope of finding a suitable CD-lattice which is not a finite group interval lattice. For a given finite nonabelian simple group \(L\), let \(\mathcal T(L)\) be the set of triples \(\tau=(H,N_H,I_H)\) with \(H\) a finite group, \(I_H\trianglelefteq N_H\leq H\) and \(F^*(N_H/I_H)\cong L\). For \(\tau\in\mathcal T(L)\), write \(N_0\) for the preimage in \(N_H\) of \(\text{Inn}(L)\) and consider the set \(\mathcal P\) of the pairs \((V,K)\) where \(V\) is an \(N_H\)-invariant subgroup of \(H\) with \(V\cap N_H=I_H\), \(K\in\mathcal O_{N_H(V)}(VN_H)\) and \(N_0V/V=F^*(K/V)\). Partially order \(\mathcal P\) by \((V_1,K_1)\leq(V_2,K_2)\) if \(V_2\leq V_1\) and \(K_2\leq K_1\): the signalizer lattice \(\Lambda(\tau)\) is the poset obtained by adjoining a least element to \(\mathcal P\). The author proves that if \(H\) admits a CD-signalizer, then the structure of \(H\) is highly restricted. More precisely, let \(\Lambda\) be a CD-lattice and assume \(\tau(H,N_H,I_H)\in\mathcal T(L)\), \(\Lambda\) is isomorphic to \(\Lambda(\tau)\) or its dual, and \(H\) is minimal subject to this constraint: then \(F^*(H)\) is the direct product of nonabelian simple subgroups permuted transitively by \(H\). A stronger result is obtained if \(\Lambda\) is a \(D\Delta(m_1,\dots,m_t)\)-lattice (i.e., the lattice \(\Lambda'\) obtained by removing from \(\Lambda\) the greatest and the least element has \(t\) connected components \(C_1,\dots,C_t\) and \(C_i\cong \Delta(m_i)'\), where \(\Delta(m_i)\) is the lattice of all subsets of an \(m_i\)-set): if \(t>1\), \(m_i>2\) and \(\Lambda=D\Delta(m_1,\dots,m_t)\) is a finite group interval lattice, then there exists an almost simple group \(G\) such that either \(\Lambda\cong\mathcal O_G(H)\) for some \(H\leq G\) or there exists \(L\) and \(\gamma=(G,N_G,I_G)\in\mathcal T(L)\) with the following properties: if \(\mathcal W_1\) is the set of the \(N_H\)-invariant subgroups \(W\) of \(H\) with \(W\cap N_H=I_H\) and \(W\leq F^*(H)I_H\), then \(G=F^*(G)N_G=\langle\mathcal W_1,N_H\rangle\) and \(\Lambda\) is isomorphic to the poset obtained by adjoining a greatest element to \(\mathcal W_1\).