The Treewidth Boundedness Problem for an Inductive Separation Logic of Relations
Marius Bozga, Lucas Bueri, Radu Iosif, Florian Zuleger
TL;DR
This paper studies the treewidth boundedness problem (TWB) for a generalization of Separation Logic to relational signatures (SLR). It proves undecidability of TWB for first-order logic while establishing decidability for TWB in SLR via a two-step approach: (i) solvability for expandable SIDs with nullary predicates using canonical and RGB-conformant color abstractions, and (ii) a reduction from arbitrary SIDs to expandable ones through a tree-automata framework that manipulates persistent variables. The core contributions are the decidability result for TWB in expandable SLR, the robust automata-based reduction technique, and a clear separation between the decidable SLR fragment and the FO/MSO boundary. The results enable automatic reasoning about the treewidth of graph-like models defined by SLR formulas and support potential decidable entailment fragments, with practical impact on verification and analysis of relational data structures. The work also clarifies the limits of expressivity by showing FO undecidability, strengthening the theoretical understanding of the MSO/ treewidth landscape in logic-based verification.
Abstract
The treewidth boundedness problem for a logic asks for the existence of an upper bound on the treewidth of the models of a given formula in that logic. This problem is found to be undecidable for first order logic. We consider a generalization of Separation Logic over relational signatures, interpreted over standard relational structures, and describe an algorithm for the treewidth boundedness problem in the context of this logic.
