Type 1 representable lattices of dimension at most 4 (Q412009)

Elaborating on earlier work of Schützenberger, \textit{B. Jónsson} [Math. Scand. 1, 193--206 (1953; Zbl 0053.21304)] introduced the Arguesian identity and showed it to be valid in all lattices of pairwise permuting equivalence relations, whence in particular in all lattices which embed into the subgroup lattice \(L(G)\) of some abelian group \(G\). Of course, the latter includes the class \(\mathcal V\) of lattices which embed into a direct product of subspace lattices of vector spaces. In [Math. Scand. 7, 133--145 (1959; Zbl 0095.34602)], \textit{B. Jónsson} proved that \(\mathcal J_4 \subseteq \mathcal V\), where \(\mathcal J_4\) denotes the class of Arguesian lattices of height \(\leq 4\). Here, the crucial case is where \(L\) is the union of two simple height \(3\) intervals -- with great impact on the theory of modular lattices and their applications in universal algebra. Based on Jónsson's structural analysis (which anticipated later decompositons as `glued sums'), the authors construct, for certain types of finite \(L \in \mathcal J_4\), a finite abelian group \(G\) such that \(L\) embeds into \(L(G)\). For finite \(L \in \mathcal J_4\) such embedding is known to be possible if and only if any simple height-\(3\) interval of \(L\) is coordinatized over some field \(\mathbb Z/p\mathbb Z\) and if, in case such one exists, any height-\(2\) interval has at most \(p+3\) elements. The paper under review has finiteness and non-existence of simple height-\(3\) intervals (i.e. 2-distributivity) as a tacit assumption. In Theorem 1.1. one has to replace \(\mathcal A\) by \(\mathcal L(\mathbb Z)\) and \(\mathcal N\) by \(\mathcal N(\text{rep})\).