Gödel algebras: interactive dualities and their applications. (Q494623)

The theory presented in this paper gives information about a technique for deriving certain new natural dualities for any variety of Heyting algebras generated by a finite chain. An algebra \(A=(A;\vee,\wedge,\to,0,1)\) is a \textit{Heyting algebra} if \((A;\vee,\wedge,0,1)\) is a bounded distributive lattice and \(a\wedge b\leq c\) iff \(a\leq b\to c\). The Heyting algebras form a variety. We shall denote this variety by \(\mathfrak H\). The variety \(\mathfrak G\) of Gödel algebras (also known as Gödel/Dummett algebras, \(L\)-algebras, pre-linear Heyting algebras or relative Stone algebras with 1) is the subvariety of \(\mathfrak H\) consisting of those algebras which satisfy the identity \((a\to b)\vee(b\to a)=1\). It is known that every subdirectly irreducible Gödel algebra is a chain. Denote the \(n\)-element chain, \(0=c_0<c_1<\cdots<c_{n-1}=1\), as Gödel algebra \(C_n\). Then the lattice of subvarieties of the variety \(\mathfrak G\) is the chain \(\mathfrak G_1\subseteq\mathfrak G_2\subseteq\cdots\subseteq\mathfrak G\), where \(\mathfrak G_n\) is generated by \(C_n\). \textit{B. A. Davey} [Trans. Am. Math. Soc. 221, 119-146 (1976; Zbl 0319.06007)] developed a duality theory for each variety \(\mathfrak G_n\). He showed that the dual of an algebra from \(\mathfrak G_n\) is a Boolean space endowed with a continuous action of the endomorphism monoid of \(C_n\). For \(n\geq 4\) the duality is not full. The authors discuss the role of a family containing both endomorphisms and partial endomorphisms of \(C_n\) for \(n\geq 4\). They find good candidates for inclusion in an alter ego for \(C_n\) translation between the associated natural duality for \(\mathfrak G_n\) and Priestley/Esakia duality: \((C_n;G,H,\tau)\), where \(G\subseteq\text{End\,}C_n\), \(H\subseteq\text{End}_pC_n-\text{End\,}C_n\) and \(\tau\) is a discrete topology. A very important role are playing here the \textit{interactive dualities}. In addition, the authors prove that for each \(n\geq 4\) only one of our \(2^{n-3}\times(n-2)\) dualities is full. New results on amalgamation are also obtained.