Various interplays between relation and cylindric algebras

Abstract

Using model theoretic techniques that proved that the class of $n$ neat reducts of $m$ dimensional cylindric algebras, $\Nr_n\CA_m$, is not elementary, we prove the same result for $\Ra\CA_k$, $k\geq 5$, and we show that $\Ra\CA_k\subset S_c\Ra\CA_k$ for all $k\geq 5$. Conversely, using the rainbow construction for cylindric algebra, we show that several classes of algebras, related to the class $\Nr_n\CA_m$, $n$ finite and $m$ arbitrary, are not elementary. Our results apply to many cylindric-like algebras, including Pinter's substitution algebras and Halmos' polyadic algebras with and without equality. The techniques used are essentially those used by Hirsch and Hodkinson, and later by Hirsch in \cite{hh} and \cite{r}. In fact, the main result in \cite{hh} follows from our more general construction. Finally we blow up a little the {\it blow up and blur construction} of Andr\'eka nd N\'emeti, showing that various constructions of weakly representable atom structures that are not strongly representable, can be formalized in our blown up, blow up and blur construction, both for relation and cylindric algebras. Two open problems are discussed, in some detail, proposing ideas. One is whether class of subneat reducts are closed under completions, the other is whether there exists a weakly representable $\omega$ dimensional atom structure, that is not strongly representable. For the latter we propose a lifting argument, due to Monk, applied to what we call anti-Monk algebras (the algebras, constructed by Hirsch and Hodkinson, are atomic, and their atom structure is stongly representable.)