Cube term blockers without finiteness (Q1689289)

A subset of \({A^d}\) is a \(d\)-ary cross on a set \(A\) iff it is a union of blocks with one restricted coordinate: \(\mathrm{cross}({U_0},\dots,{U_{d - 1}}) = {B_0} \cup \dots \cup {B_{d - 1}}\), where \({B_i} = A \times \dots \times A \times {U_i} \times A \times \dots \times A\), \({U_i} \subseteq A\), \({U_i} \neq \emptyset \). A term \(c\) of a variety \(\mathcal{V} \) is called \(d\)-cube term if \(c\) yields \({\mathbf{y}} = (y,\dots,y)\) when applied to elements of \({\{ x,y\} ^d} - \{ {\mathbf{y}}\} \). It is proved that an idempotent variety has a \(d\)-cube term iff its free algebra on two generators has no \(d\)-ary compatible cross. If an idempotent variety \(\mathcal{V} \) with operation arities \({n_1},\dots,{n_k}\), \({n_i} \geqslant 2\) has a \(d\)-cube term then the sharp upper bound for the minimal \(d\) is \(1 + \sum_i {({n_i}} - 1)\). If \(\mathcal{V} \) does not have a cube term of this size, then it does not have a cube term at all. Thus if the signature of \(\mathcal{V} \) consists of a single binary operation, then either \(\mathcal{V}\) has a Maltsev term or it has no cube term at all.