Compatibility of Shelah and Stupps and Muchniks iteration with fragments of monadic second order logic

We investigate the relation between the theory of the iterations in the sense of Shelah-Stupp and of Muchnik, resp., and the theory of the base structure for several logics. These logics are obtained from the restriction of set quantification in monadic second order logic to certain subsets like, e.g., finite sets, chains, and finite unions of chains. We show that these theories of the Shelah-Stupp iteration can be reduced to corresponding theories of the base structure. This fails for Muchnik’s iteration.

💡 Research Summary

The paper investigates how the logical theories of two well‑known tree‑like constructions—Shelah‑Stupp iteration and Muchnik iteration—relate to the theory of the underlying base structure when the quantification over sets is restricted in various ways. The authors focus on fragments of monadic second‑order logic (MSO) obtained by limiting set variables to finite sets (MSO_w), to chains (MSO_ch), to finite unions of chains (MSO_mch), or to closed sets in a natural topology (MSO_closed).

Background.
Rabin’s tree theorem shows that the full MSO theory of the complete binary tree is decidable. This result has been generalized by Shelah and Stupp, who introduced an iteration that builds a tree whose nodes are all finite words over the universe of a given structure A, ordered by the prefix relation, and where each node inherits the relations of A. Muchnik later added a unary “clone” predicate that holds exactly on words whose last two letters coincide. This predicate enables the unfolding of rooted graphs inside the iteration and underlies the Caucal hierarchy of infinite graphs with decidable MSO theory.

Logical fragments.
The paper defines four restricted versions of MSO:

• MSO_w – set quantifiers range only over finite subsets.
• MSO_ch – set quantifiers range over chains (linearly ordered subsets).
• MSO_mch – set quantifiers range over finite unions of chains (multichains).
• MSO_closed – set quantifiers range over closed subsets of a topology induced by the structure.

These fragments are motivated by automata‑theoretic techniques and by the desire to capture decidable fragments of MSO on infinite structures.

Main positive results for Shelah‑Stupp iteration.

1. MSO_closed‑compatibility. By adapting Rabin’s basis theorem, the authors show that any MSO_closed formula true in the Shelah‑Stupp iteration can be witnessed by a regular language of words. Consequently, the MSO_closed theory of the iteration reduces uniformly to the full MSO theory of the base structure A.

2. FO‑compatibility. Building on earlier work, they confirm that the first‑order (FO) theory of the infinite Shelah‑Stupp iteration is uniformly reducible to the FO theory of A. The reduction uses the fact that the iteration is essentially a disjoint union of copies of A indexed by prefixes, and that the mapping which replaces a common prefix by another is an isomorphism.

3. MSO_w‑compatibility. The crucial observation is that finiteness of a set of nodes in the iteration can be expressed in MSO_mch (quantification over finite unions of chains). Since MSO_w ⊆ MSO_mch, any MSO_w property of the iteration can first be translated into an MSO_mch property, and then, using the basis theorem, into a property of A. The authors give an explicit construction: a regular language accepted by a deterministic finite automaton is represented by its transition matrix, which consists of a fixed finite number of sets definable in A. This yields a uniform reduction from MSO_w(Shelah‑Stupp) to MSO(A).

Negative result for Muchnik iteration.
The clone predicate dramatically increases expressive power. The authors prove that the infinite Muchnik iteration is not MSO_w‑compatible. The proof proceeds by encoding an arbitrary set M ⊆ ℕ into a tree A_M such that A_M is MSO_w‑equivalent to a fixed tree T_ω (which has a decidable MSO_w theory). In the Muchnik iteration of A_M, the existence of an infinite branch becomes first‑order definable (thanks to the clone predicate), while MSO_w cannot express this property. Hence, if MSO_w of the iteration reduced to MSO_w of the base, one could decide arbitrary subsets of ℕ, contradicting known undecidability results. A later simplification by Aleks Bes, using automatic structures, confirms the same separation.

Methodological contributions.

• The paper refines the use of Hintikka‑style formulas to bound the number of MSO_mch equivalence classes, which is essential for the uniform reduction.
• It introduces a “short‑word” lemma showing that any consistent MSO_mch property of a single node can be witnessed by a word of bounded length, a key step in the basis theorem adaptation.
• The authors systematically employ product constructions and infinite products of τ‑structures to model concatenations of words and to control the logical equivalence across different prefixes.

Implications.

• For the Caucal hierarchy, the result clarifies that while the hierarchy enjoys decidable full MSO, it does not preserve decidability for weaker fragments such as MSO_w when the clone predicate is present.
• In the theory of automatic structures, the positive results show that Shelah‑Stupp iteration can be safely used to build more complex structures without leaving decidable fragments of MSO, whereas Muchnik iteration must be handled with care.
• The separation between the two iterations highlights how a seemingly modest unary predicate can shift a structure from being compatible with weak monadic logics to being incompatible, informing future design of logical extensions for infinite-state systems.

Conclusion.
The paper establishes a clear dichotomy: Shelah‑Stupp iteration is uniformly compatible with FO, full MSO, and all considered weak MSO fragments, while Muchnik iteration fails to be compatible with MSO_w (and consequently with other weak fragments). This delineates the exact logical power added by the clone predicate and provides a solid foundation for further investigations into logical reductions, decidability, and the expressive limits of iterated tree constructions.