A Dichotomy Theorem for Ordinal Ranks in MSO

We focus on formulae $\exists X., φ(\vec{Y}, X)$ of monadic second-order logic over the full binary tree, such that the witness $X$ is a well-founded set. The ordinal rank $\mathrm{rank}(X) < ω_1$ of such a set $X$ measures its depth and branching structure. We search for the least upper bound for these ranks, and discover the following dichotomy depending on the formula $φ$. Let $\mathrm{rank}(φ)$ be the minimal ordinal such that, whenever an instance $\vec{Y}$ satisfies the formula, there is a witness $X$ with $\mathrm{rank}(X) \leq \mathrm{rank}(φ)$. Then $\mathrm{rank}(φ)$ is either strictly smaller than $ω^2$ or it reaches the maximal possible value $ω_1$. Moreover, it is decidable which of the cases holds. The result has potential for applications in a variety of ordinal-related problems, in particular it entails a result about the closure ordinal of a fixed-point formula.

💡 Research Summary

The paper investigates a quantitative aspect of monadic second‑order logic (MSO) over the full binary tree. For formulas of the form ∃X φ( Y⃗ ,X ) the authors require that any witness X be a well‑founded set of nodes, i.e., it contains no infinite ascending chain with respect to the tree’s descendant order. Such a set admits an ordinal rank rank(X) < ω₁, defined inductively as the supremum of the ranks of its children plus one.

The central notion introduced is rank(φ), the least ordinal that bounds the rank of a witness for every possible valuation of the free set variables Y⃗ . Formally,
 rank(φ) = sup_{Y⃗ } min{ rank(X) | X satisfies φ(Y⃗ ,X) }.
Since the supremum ranges over countable ordinals, rank(φ) ≤ ω₁.

The main theorem (Theorem 1.1) establishes a sharp dichotomy: for any such formula, rank(φ) is either strictly below ω² (equivalently, there exists a natural number N with rank(φ) < ω·N) or it equals the maximal possible value ω₁. Moreover, it is decidable which case holds, and in the former case an explicit bound N can be computed effectively. The authors also prove that the property “rank(φ) < ω²” is not expressible in MSO itself, contrasting with known results about cardinality quantifiers.

To obtain the dichotomy, the authors employ a game‑theoretic method. They construct a finite‑arena, perfect‑information game between two players, ∃ and ∀, whose winning condition is an ω‑regular language. By the Büchi–Landweber theorem, such games are determined and the winner can be computed algorithmically. The game is designed so that a winning strategy for ∃ corresponds to the existence of witnesses of arbitrarily high rank, forcing rank(φ)=ω₁, while a winning strategy for ∀ produces a uniform bound N ensuring rank(φ)<ω·N. The construction carefully encodes the structure of potential witnesses into the game’s moves, guaranteeing the two implications.

Beyond the core result, the paper explores several applications. In the µ‑calculus, the closure ordinal of a formula μX.F(X) is precisely the rank of the associated MSO definition of F. The dichotomy therefore answers a question posed by Czarnecki (2010) about the existence of µ‑calculus formulas with closure ordinals ≥ ω²: such formulas exist only when the closure ordinal is actually ω₁, i.e., only in uncountable models. The authors also extend the analysis to vectorial fixed points μ ⃗X. ⃗F( ⃗X ), showing an analogous dichotomy (Theorem 10.2).

The work situates itself among several related lines of research. It refines earlier results on the expressive power of MSO with cardinality quantifiers, shows that certain ordinal‑related properties lie beyond MSO’s reach, and connects to known dichotomies for tree automata (e.g., languages recognized by nondeterministic Büchi automata are either Π₁¹‑complete or lie low in the Borel hierarchy). Comparisons with automatic structures reveal that while automatic ordinals can reach ω^{ω} or higher, the MSO‑definable setting imposes the stricter bound ω².

In summary, the paper delivers a robust, decidable classification of the ordinal complexity of MSO‑definable witnesses over the binary tree, introduces a novel game‑based proof technique, and leverages the result to obtain new insights into closure ordinals of µ‑calculus formulas and related automata‑theoretic phenomena.