arXiv Analytics

Sign in

arXiv:1408.3282 [math.LO]AbstractReferencesReviewsResources

Atom-canonicity and complete representations for cylindric-like algebras, and omitting types for the clque guarded fragment of first order logic

Tarek Sayed Ahmed

Published 2014-06-25Version 1

Fix a finite ordinal n>2. We show that there exists an atomic, simple and countable representable CA_n, such that its minimal completion is outside SNr_nCA_{n+3}. Hence, for any finite k\geq 3, the variety SNr_nCA_{n+k} is not atom-canonical, so that the variety of CA_n's having n+k-flat representations is not atom-canonical, too. We show, for finite k\geq 3, that S_cNr_nCA_{n+k} is not elementary, hence the class of CA_n's having complete n+3-smooth representations is not elementary. We obtain analogous results by replacing flat and smooth, respectively, by (the weaker notion of) square; this give a stronger result in both cases and here we can allow k to be infinite. Our results are proved using rainbow constructions for CA's. We lift the negative result on atom-canonicity to the transfinite. We also show that for any ordinal \alpha\geq \omega, for any finite k\geq 1, and for any r\in \omega, there exists an atomic algebra A_r\in SNr_\alphaCA_{\alpha+k}\sim SNr_nCA_{\alpha+k+1}, such that \Pi_{r/U} A_r\in RCA_{\alpha} where U is any non--principal ultrafilter on \omega. Reaping the harvest of our algebraic results we investigate a plethora of omitting types theorems for variants of first logic including its finite variable fragments and its packed fragment.

Comments: arXiv admin note: text overlap with arXiv:1308.6165, arXiv:1307.1016
Categories: math.LO
Related articles: Most relevant | Search more
arXiv:1807.00690 [math.LO] (Published 2018-06-29)
First order logic without equality on relativized semantics
arXiv:1605.03501 [math.LO] (Published 2016-05-11)
Is Leibnizian calculus embeddable in first order logic?
arXiv:2108.03846 [math.LO] (Published 2021-08-09)
On failure of 0-1 laws