Generalizing semidistributivity (Q1311423): Difference between revisions

A lattice \(L\) is \(k\)-join distributive if \(a\vee b_ 1= a\vee b_ 2=\cdots= a\vee b_{k+ 1}\) implies that \(a\vee b_ 1= \bigvee_{i\neq j} a\vee (b_ i\wedge b_ j)\). The dual concept is \(k\)-meet distributivity. For \(k= 1\), we get the usual concept of join-distributive (and meet-distributive) lattices. Every finite lattice is \(k\)-join distributive for sufficiently large \(k\). \textit{Alan Day's} interval doubling construction has been extended by many authors to convex sublattices. He proved [Can. J. Math. 31, 69-78 (1979; Zbl 0432.06007)] that \(\ell[C]\), the class of lattices obtained from the one-element lattice by repeating doubling of convex sublattices, can be characterized as, so-called, congruence normal lattices. If we further impose the condition that the convex sets be upper pseudo-intervals (convex sets with a common greatest element), then the characterization adds the meet semi-distributive law. The author verifies a conjecture of Day: the class of lattices obtained from the one-element lattice by repeated doubling of convex sets with at most \(k\) minimal and \(\ell\) maximal elements is a pseudo-variety. This class can be described using the generalized semidistributive laws.