On a new construction of pseudocomplemented semilattices (Q2238023): Difference between revisions

A pseudocomplemented semilattice is an algebraic structure \((L; \wedge, {}^*, 0, 1)\) where \((L; \wedge, 0, 1)\) is a bounded meet semilattice and, for any \(a\in L\), \(a^*\) is the largest element \(x\) satisfying \(a\wedge x = 0\). An element \(a\in L\) is closed if \(a=a^{**}\). The set \(B(L)\) of closed elements of \(L\) is subalgebra of \(L\) and becomes a Boolean algebra \((B(L); +, \wedge, {}^*, 0, 1)\) if we add the conjugate of the meet operation, namely \(x+y:=(x^*\wedge y^*)^*\). The function \(x\mapsto x^{**}\) is a retraction of \(L\) onto its Boolean algebra of closed elements, and the kernel of this endomorphism is the Glivenko-Frink congruence. The Glivenko-Frink congruence class \(\Gamma_c\) of a closed element \(c\) is a subsemilattice with unit element \(c\). Let \(\mathfrak G_L\) be the set of all subsemilattices \(\Gamma_c, c\in L\). If \(c\geq d\) are both closed, then the ``translation map'' \(\lambda_{c,d}(x) = x\wedge d\) is a semilattice homomorphism from \(\Gamma_c\) to \(\Gamma_d\) which preserves unit elements (meaning that \(\lambda_{c,d}(c)=d\)). Note that \(\Gamma_0=\{0\}\) and \(\lambda_{c,c}(x)\) is the identity function on \(\Gamma_c\). If \(L\) is a pseudocomplemented semilattice, then the triple \([B(L), {\mathfrak G}, \lambda_{a,b}]\) is called the \emph{associated full triple} of \(L\). Everything in the previous paragraph can be reversed to produce a decomposition theorem for pseudocomplemented semilattices. The necessary data are: a Boolean algebra \(B\), a \(B\)-indexed family \(\mathfrak G\) of pairwise disjoint unital semilattices, \(\Gamma_c, c\in B\), and a family of unital semilattice homomorphisms \(\varphi_{c,d}\colon \Gamma_c\to \Gamma_d\), \(c\geq d\). We always assume that \(\Gamma_0 = \{0\}\) and \(\varphi_{c,c}\) is the identity function on \(\Gamma_c\) for all \(c\in B\). If \([B, {\mathfrak G}, \varphi_{a,b}]\) is a data sequence satisfying these conditions, then it is called an \emph{abstract full triple}. The authors prove that any associated full triple is an abstract full triple (Proposition 3.1), and that any abstract full triple is the associated full triple of a uniquely determined pseudocomplemented semilattice (Corollary 5.2). They also study homomorphisms and subalgebras of pseudocomplemented semilattices in terms of their associated triples.