---
title: 'C₂(omega): A Countable Second-Order Inner Model'
url: https://www.emergentmind.com/topics/c2-omega
type: topic
---

# C₂(omega): A Countable Second-Order Inner Model

Searching arXiv for the exact notation and primary paper.
arxiv_search(query="2508.17672", max_results=5, sort_by="submittedDate")

$C_2(\omega)$ is the inner model $C(\mathcal{L}^2_\omega)$ based on a fragment of second-order logic in which second-order variables range over countable subsets of the domain. In this construction, definability is strengthened beyond first-order $L$-style definability but remains weaker than full second-order definability, because the quantifiers range only over countable sets in the ambient universe $V$. The resulting model is transitive, contains all ordinals, is contained in the Chang model $C_{\omega_1\omega}$, and, under strong large-cardinal hypotheses, has generic absoluteness and substantial inner large-cardinal content [2508.17672].

## 1. Logical basis

The logic underlying $C_2(\omega)$ is $\mathcal{L}^2_\omega$. Its first-order variables $x,y,z$ range over the domain $D$ of a structure $M$, while its second-order variables $X,Y,R,\dots$ range over countable subsets of $D^k$ or countable subsets of $D$. The crucial semantic feature is externality: in a structure $(M,\in)$, the quantifiers $\exists X$ and $\forall X$ range over sets that are countable in $V$, not merely countable in $M$, and not necessarily elements of $M$ itself. Thus $M\models \exists X\,\varphi(X,a)$ means that there is a set $X\subseteq D^k$ that is countable in $V$ such that $M\models \varphi(X,a)$.

This makes $\mathcal{L}^2_\omega$ a fragment of second-order logic with bound second-order quantification restricted to countable sets or relations. The paper describes this logic as incompact and notoriously non-axiomatizable, but still robust enough to support a $C(\mathcal{L})$-style inner-model construction. Definability in $C_2(\omega)$ is by $\mathcal{L}^2_\omega$-formulas over the current stage, using parameters from that stage. Because second-order variables range over countable subsets in $V$, definability can reach out to countable pieces of the ambient universe even when those pieces are not internal to the stage under consideration [2508.17672].

A basic comparison point is full second-order logic. The paper states that $C(\mathcal{L}^2)=\mathrm{HOD}$, while the restriction to countable second-order quantification places $C_2(\omega)$ strictly below $\mathrm{HOD}$ in general. This difference is central to the model’s behavior: $C_2(\omega)$ is intended to capture a level of definability sensitive to externally countable structure without collapsing to full ordinal definability.

## 2. Stage-by-stage construction

Conceptually, $C_2(\omega)$ is obtained by replacing the definability step of Gödel’s $L$-hierarchy with definability in $\mathcal{L}^2_\omega$. To ensure the Axiom of Choice, the construction uses the “truth-folded” definition of $C(\mathcal{L})$: at each stage one folds in a truth predicate for $\mathcal{L}^2_\omega$ over the current level.

Formally, the hierarchy $(J'_\alpha)_{\alpha\in\mathrm{Ord}}$ is defined by double induction. One introduces truth-codes $T\subseteq \mathrm{Ord}\times \mathrm{Form}\times \mathrm{Params}$ such that
$$(\alpha,\varphi,x)\in T \iff (J'_\alpha,\in,\alpha)\models \varphi(x),$$
where $\varphi\in\mathcal{L}^2_\omega$ and $x\in J'_\alpha$. The stages are then defined by
$$J'_0=\varnothing,$$
$$J'_{\omega\nu}=\bigcup_{\alpha<\nu}J'_{\omega\alpha}\quad\text{for limit }\nu,$$
and
$$J'_{\alpha+\omega}=\mathrm{rud}_T\big(J'_\alpha\cup\{J'_\alpha\}\big),$$
where $\mathrm{rud}_T$ is the rudimentary closure augmented to allow the operation $x\mapsto x\cap T$ and interpretations of $\mathcal{L}^2_\omega$-truth at stage $\alpha$. Finally,
$$C_2(\omega)=\bigcup_{\alpha\in\mathrm{Ord}}J'_\alpha.$$

The paper also states that one may define $C_2(\omega)$ using countable sequences instead of countable subsets; the resulting model is the same. With the truth-folded construction, $C_2(\omega)$ satisfies $\mathrm{ZF}$ and also $\mathrm{AC}$. By contrast, the older definition may fail to imply $\mathrm{AC}$ for $\mathcal{L}^2_\omega$, because truth in $\mathcal{L}^2_\omega$ is not guaranteed to be internally adequate [2508.17672].

## 3. Structural properties

Several basic structural properties are established. $C_2(\omega)$ is transitive, contains all ordinals, and is definable as a subclass of $V$. It is also contained in the Chang model $C_{\omega_1\omega}$, and repeating the construction inside $C_{\omega_1\omega}$ yields the same $C_2(\omega)$. This containment is important because it identifies $C_2(\omega)$ as an inner model built from countable information but still strictly controlled by a canonical closure condition.

A cardinality bound is proved: for every $\kappa\ge 2$,
$$|\mathcal{P}(\kappa)\cap C_2(\omega)|\le (\kappa^\omega)^+.$$
The proof described in the paper uses a chain of $\mathcal{L}^2_\omega$-elementary submodels of $H_\mu$ together with a collapse argument. This places a strong restriction on how much of the power set of a cardinal can be captured by $C_2(\omega)$.

The model is compatible with the failure of the Continuum Hypothesis. The paper proves
$$\mathrm{Con}(\mathrm{ZF}) \Rightarrow \mathrm{Con}(C_2(\omega)\models \neg\mathrm{CH}).$$
The argument uses Harrington’s long projective well-ordering to produce a model of $\neg\mathrm{CH}$ with a projective well-order $<^*$ and then shows that $C_2(\omega)$ contains all reals of the ambient universe in that setting. This demonstrates that $C_2(\omega)$ is not tied to the combinatorics of $L$ and can accommodate non-$\mathrm{CH}$ behavior [2508.17672].

## 4. Comparison with $C(aa)$ and $C_2(\omega,aa)$

A central theme of the paper is the comparison between $C_2(\omega)$ and the stationary-logic inner model $C(aa)$. In stationary logic, the generalized quantifier $aa$ binds variables over $[M]^\omega$, and the statement $aa\,s\,\varphi(s)$ means that $\{s\in [M]^\omega : M\models \varphi(s)\}$ is a club subset of $[M]^\omega$. The hybrid model $C_2(\omega,aa)$ is obtained by combining $\mathcal{L}^2_\omega$ with the $aa$-quantifier in the same truth-folded style.

The paper records the trivial inclusions
$$C^*\subseteq C_2(\omega)\subseteq C_2(\omega,aa)\subseteq V,$$
and
$$C^*\subseteq C(aa)\subseteq C_2(\omega,aa).$$
Beyond these inclusions, the relationship between $C_2(\omega)$ and $C(aa)$ is delicate. In ZFC alone, one cannot prove that $C_2(\omega)$ is “bigger” than $C(aa)$ in any absolute sense. Indeed, the paper gives a consistency result showing
$$\mathrm{Con}(\mathrm{ZF}) \Rightarrow \mathrm{Con}(C_2(\omega)\subsetneq C(aa)).$$
The proof starts from $V=L$, adds a Cohen real by homogeneous c.c.c. forcing so that $C_2(\omega)$ remains $L$, and then codes that real into a stationary pattern on $\omega_1$ without adding countable sets; in the final model the coded real belongs to $C(aa)$, so $C_2(\omega)=L\subsetneq C(aa)$.

At the same time, under stronger large-cardinal assumptions the comparison shifts. Assuming a proper class of Woodin cardinals, the paper states that $C_2(\omega)\not\subseteq C(aa)$ and that for every $n\in\omega$, $C_2(\omega)$ contains an inner model with $n$ Woodin cardinals, whereas under the same assumption $C(aa)$ contains no inner model with a Woodin cardinal. It also proves that all reals of $C(aa)$ are in $C_2(\omega)$ under a proper class of Woodin cardinals, and under a stronger hypothesis denoted “$^{++}$” every subset of $\omega_1$ in $C(aa)$ is in $C_2(\omega)$ [2508.17672].

## 5. Large cardinals inside and around $C_2(\omega)$

The large-cardinal behavior of $C_2(\omega)$ is one of the paper’s main results. Assuming a proper class of Woodin cardinals, $C_2(\omega)$ contains, for every finite $n$, an inner model with $n$ Woodin cardinals. The argument uses the fact that for each $n$, $M_n^\sharp$ is a $\Pi^1_{n+2}$-singleton and is therefore definable in $C_2(\omega)$; iterating the top measure inside $C_2(\omega)$ then yields the desired inner models.

The paper also proves a reflection result for $\omega_1^V$. Assuming a Woodin limit of Woodin cardinals, $\omega_1^V$ is strongly Mahlo in $C_2(\omega)$. The proof uses the countable stationary tower $Q_{<\delta}$ at a Woodin limit $\delta$ to obtain an embedding $j:V\to M$ with ${}^\omega M\subseteq M$, then argues by elementarity that $\omega_1^V$ is inaccessible in $C_2(\omega)$ and finally strongly Mahlo there.

An even stronger layer of structure appears under a proper class of Woodin limits of Woodin cardinals. The paper invokes Woodin’s principle $CM^+$ and proves that $C_2(\omega)$ satisfies Club Determinacy: every stage $(J'_\alpha,\in,\alpha)$ decides every $\mathcal{L}^2_\omega(aa)$-formula with parameters from $J'_\alpha$ and countable parameter sequences. From Club Determinacy it deduces that every regular cardinal of $V$ is measurable in $C_2(\omega)$. The normal ultrafilter is defined by declaring $X\in\mathcal{U}$ iff, at a sufficiently high stage, the corresponding structure satisfies that $\sup(s\cap\kappa)$ lies in $X$; Club Determinacy forces either $X$ or its complement into $\mathcal{U}$ [2508.17672].

## 6. Absoluteness, forcing, and surrounding inner-model context

Under a proper class of Woodin limits of Woodin cardinals, the theory of $C_2(\omega)$ is generically absolute: forcing cannot change it. More precisely, for any forcing $\mathbb{P}$ and any generic $G$, one has
$$(C_2(\omega))^V \equiv (C_2(\omega))^{V[G]}.$$
The paper derives this from Woodin-style absoluteness, using collapse models $V^{\mathrm{Coll}(\omega,<\delta)}$ at Woodin limits of Woodin cardinals together with homogeneity. This means that, under the stated hypothesis, the first-order theory of $C_2(\omega)$ is fixed across forcing extensions.

The broader context includes two further inner-model comparisons. First, under $V=L^\mu$, the paper states that $V=C(aa)$, while the modified construction “$2,+$” equals $M_\omega$; from this it concludes that $C_2(\omega)\subset C(aa)$. Second, the paper discusses $\mathrm{HOD}_1$, a variant of $\mathrm{HOD}$ arising from existential second-order definability, and states that the assertion $\mathrm{HOD}_1=\mathrm{HOD}$ is independent of $\mathrm{ZFC}$ even when one adds the existence of supercompact cardinals. One consistency direction uses a Menas model of $V=\mathrm{HOD}$ with a supercompact cardinal; the other starts from $V=\mathrm{HOD}+\mathrm{GCH}$ with a supercompact cardinal and uses homogeneous Easton-support iterations and Souslin tree forcing to keep $\Sigma^1_1$-truth stable while changing $\mathrm{HOD}$ [2508.17672].

Taken together, these results place $C_2(\omega)$ in a distinctive position among definability-based inner models. It is weaker than full second-order $\mathrm{HOD}$, contained in the Chang model, and sensitive to externally countable structure; yet under strong large-cardinal assumptions it exhibits generic absoluteness, Club Determinacy, and inner models with finitely many Woodin cardinals. This suggests that $C_2(\omega)$ functions as a bridge between $L$-style definability hierarchies and inner models informed by countable second-order information.

Source: https://www.emergentmind.com/topics/c2-omega