Decidable Expansions of Labelled Linear Orderings
Consider a linear ordering equipped with a finite sequence of monadic predicates. If the ordering contains an interval of order type \omega or -\omega, and the monadic second-order theory of the combined structure is decidable, there exists a non-trivial expansion by a further monadic predicate that is still decidable.
š” Research Summary
The paper investigates decidability and maximality issues for monadic secondāorder (MSO) theories of labelled linear orderings, i.e., structures of the form M = (A, <, Pā,ā¦,Pā) where < is a total order on A and each Pįµ¢ is a unary predicate (a ālabelā). The central question, motivated by the classic ElgotāRabin problem, asks whether there exist structures whose MSO theory is decidable but becomes undecidable after any nonādefinable expansion. The authors show that for a very broad class of labelled linear orderings this never happens: if the underlying order (A, <) contains an interval isomorphic to Ļ (the natural numbers) or to āĻ (the reverse of Ļ) and the MSO theory of M is decidable, then one can add a new unary predicate Q, not definable in M, such that the expanded structure Mā² = (A, <, Pā,ā¦,Pā, Q) still has a decidable MSO theory. Consequently, no such M is maximal with respect to MSO decidability.
The proof proceeds in two stages. First, the authors treat the canonical cases of the natural numbers (Ļ) and the integers (ζ). Using Büchiās theorem that MSO over Ļ coincides with Ļāregular languages, they construct a uniform method to colour intervals of the order with a new predicate Q while preserving decidability. The construction relies on the notion of kātype (the set of all MSO sentences of quantifier depth ā¤āÆk true in a structure). By ensuring that all intervals used in the construction share the same kātype, the MSO theory of the expanded structure can be computed from the kātypes of the original structure.
For the general case, the authors introduce a convex equivalence relation ā¼ on A that is definable in M and partitions A into convex blocks. Each block is either finite, of type āĻ, Ļ, or ζ. The structure M can then be viewed as an ordered sum of the substructures induced on these blocks, together with the quotient order M/ā¼. The crucial tool is Shelahās composition theorem for ordered sums: the kātype of a sum is effectively determined by the kātypes of its summands. Applying the construction from the first stage to each block yields a predicate Q that is defined blockāwise but globally nonādefinable in M. Because the kātype of each blockās expansion is computable from its original kātype, the overall MSO theory of Mā² reduces to the MSO theories of the blocks and of the quotient, all of which are decidable by hypothesis. Hence MSO(Mā²) is decidable.
The paper also discusses why this result does not follow from earlier criteria based on Gaifman graphs (as in
Comments & Academic Discussion
Loading comments...
Leave a Comment