Tight MAD Families in Set Theory

• Tight MAD families are maximal almost disjoint collections of infinite subsets of natural numbers with a strengthened selection property ensuring every countable collection outside the ideal meets some family member in an infinite set.
• Shelah’s creature forcing constructs generic extensions that strongly preserve tightness, maintaining the structure of MAD families without adding dominating reals.
• Models with tight MAD families achieve precise cardinal characteristics (e.g., ℵ1 = 𝔞 < 𝔰 = ℵ2), enabling definable wellorders and minimal projective complexity.

A tight maximal almost disjoint (MAD) family is a refinement of the classical notion of a MAD family, incorporating a strengthened selection property relative to countable subsets outside the ideal generated by the family. These combinatorial structures on the infinite subsets of natural numbers play a fundamental role in set theory, descriptive set theory, and the theory of definable sets of reals. Significant recent work demonstrates the preservation and definability properties of tight MAD families under forcing, and establishes the consistency of models supporting tight MAD families with specific cardinal characteristics, wellorders of the reals, and projective complexity constraints (Fischer et al., 13 Jan 2026).

1. Definitions and Fundamental Properties

Let $\omega$ denote the natural numbers, and $[\omega]^\omega$ the set of infinite subsets of $\omega$. A family $\mathcal{A} \subseteq [\omega]^\omega$ is almost disjoint if $|a \cap b| < \omega$ for all distinct $a, b \in \mathcal{A}$. It is a MAD family if $\mathcal{A}$ is maximal with respect to this property; equivalently, every infinite $b \subseteq \omega$ has infinite intersection with some $a \in \mathcal{A}$.

The ideal generated by $\mathcal{A}$ is $I(\mathcal{A}) = \{ b \subseteq \omega : \exists$ finite $$F \subseteq \mathcal{A},\ b \subseteq^* \bigcup F \}$$, where $X \subseteq^* Y$ means $X \setminus Y$ is finite. Define $$(\mathcal{A})^+ = \mathcal{P}(\omega) \setminus I(\mathcal{A})$$, the collection of sets not covered modulo finite by finitely many elements of $\mathcal{A}$.

A MAD family $\mathcal{A}$ is tight if for every countable $B \subseteq (\mathcal{A})^+$ there exists $a \in \mathcal{A}$ such that $|a \cap b| = \omega$ for all $b \in B$. Tightness immediately implies maximality, but is strictly stronger.

2. Shelah’s Creature Forcing and Forcing Preservation

Shelah's original creature forcing (denoted $\mathbb{Q}$) constructs generic extensions characterized by highly controlled combinatorics and preservation of key ground model families (Fischer et al., 13 Jan 2026). A condition in $\mathbb{Q}$ is a pair $(u, T)$, where $u \in [\omega]^{<\omega}$ and $T = \langle t_i : i \in \omega \rangle$ is a sequence of "creatures" $t_i = (s_i, h_i)$. Each creature is a finite logarithmic measure: $s_i$ is a finite subset of $\omega$, $h_i: [s_i]^{<\omega} \to \omega$, with technical subadditivity and positivity conditions.

Ordering in $\mathbb{Q}$ is designed to enable fusion: for $q \leq p$, $u'$ end-extends $u$, the illustration parts (the unions of all $s_i$) are nested, and positivity is preserved through combinatorial partitions. The fusion property guarantees any countable chain $p_0 \geq_1 p_1 \geq_2 \dots$ (with each $p_{n+1}$ agreeing with $p_n$ on the first $n$ creatures) admits a lower bound extending all at corresponding levels.

Shelah's creature forcing is proper and $\omega^\omega$-bounding; crucially, the unboundedness of creature levels prevents $\mathbb{Q}$ from adding dominating reals. Moreover, the forcing does not split any ground-model infinite subset, a fact essential for the preservation of tight MAD families.

3. Strong Preservation of Tightness

Let $\mathcal{A}$ be a tight MAD family in a ground model $V$. A poset $\mathbb{P}$ strongly preserves the tightness of $\mathcal{A}$ if for every condition $p \in \mathbb{P}$, every countable elementary submodel $M \prec H_\theta$ with $\{\mathbb{P}, \mathcal{A}, p \} \subseteq M$, and every $B \in I(\mathcal{A}) \cap M$ satisfying $$\forall Y \in I(\mathcal{A})^+ \cap M\ (|B \cap Y| = \omega)$$, there is a $q \leq p$ that is $(M, \mathbb{P})$-generic and forces that all $Z \in I(\mathcal{A})^+ \cap M[G]$ satisfy $|Z \cap B| = \omega$.

Shelah's $\mathbb{Q}$ strongly preserves tightness: any ground-model tight MAD family $\mathcal{A}$ remains tight in $V[G]$ for $G \subset \mathbb{Q}$ generic. Furthermore, countable support iterations of strongly tightness-preserving posets retain this property. The combinatorial mechanism involves careful enumeration of dense sets and fusion sequences that ensure genericity and meet all requirements for witnesses to tightness (Fischer et al., 13 Jan 2026).

4. Models with Tight MAD Families, Cardinal Characteristics, and Definability

The consistent configuration $\aleph_1 = \mathfrak{a} < \mathfrak{s} = \aleph_2$ (where $\mathfrak{a}$ is the minimal size of a MAD family, $\mathfrak{s}$ the splitting number) holds in a generic extension produced by iterating $\mathbb{Q}$ of length $\omega_2$ with countable support, starting with CH and a ground-model tight MAD family of size $\aleph_1$. At each stage, $\mathbb{Q}$ adds a real not split by any ground-model family of size $<\omega_2$, yielding $\mathfrak{s} = \aleph_2$, while preservation ensures $\mathfrak{a} = \aleph_1$ (Fischer et al., 13 Jan 2026).

By integrating Sacks-coding, club-shooting, localization, and almost disjoint coding in the iteration, and interleaving $\mathbb{Q}$-stages, one can simultaneously ensure:

• The existence of a $\Delta^1_3$-definable wellorder of the reals,
• A $\Pi^1_1$ tight MAD family of size $\aleph_1$,
• A $\Pi^1_2$ tight MAD family of size $\aleph_2$.

This model demonstrates minimal projective complexity for such definable MAD families and a definable wellorder consistent with sharply separated small cardinals.

5. Projective Complexity of Tight MAD Families

For the definable examples constructed in the aforementioned models, explicit formulas express their projective complexity:

• The coanalytic MAD family $\mathcal{A}_1$ of size $\aleph_1$ arises in $L$ via a $\Sigma_2$-recursion along $\omega_1$ and is $\Pi^1_1$ in the codes. Shoenfield absoluteness ensures the definition retains its analytic rank in the forcing extension, and the preservation arguments maintain tightness and size.
• The $\Pi^1_2$ MAD family $\mathcal{A}_2$ of size $\aleph_2$ is defined using a recursive coding $\Delta : 2^\omega \rightarrow \mathcal{P}(\omega)$ and a $\Sigma_1$-definable sequence $$\langle S_{\alpha+m} : m < \omega \rangle \subseteq \omega_1$$. The formula

$$a \in \mathcal{A}_2 \Leftrightarrow \forall \text{ countable suitable } M \, (a \in M \Rightarrow \exists \alpha < (\omega_2)^M \ M \models \forall m \in \Delta(a) \ ("S_{\alpha + m} \text{ nonstationary}") )$$

is $\Pi^1_2$. Minimality is established: no analytic or $\Sigma^1_2$ MAD family of size $>\aleph_1$ exists, so these definable examples optimize projective complexity for large MAD families (Fischer et al., 13 Jan 2026).

6. Resolution of Open Questions and Impact

The construction and preservation results answer longstanding open questions:

• The existence of a model with $\aleph_1 = \mathfrak{a} < \mathfrak{s} = \aleph_2$ and a $\Delta^1_3$ wellorder of the reals, as posed by Fischer–Friedman, is affirmed by application of $\mathbb{Q}$ and Sacks-coding.
• The existence of $\Pi^1_1$ and $\Pi^1_2$ definable tight MAD families of large size, as queried by Friedman–Zdomskyy, is achieved by appropriate coding and carefully engineered forcing iterations.
• The notion of strong preservation for tightness, introduced by Guzmán–Hrušák–Téllez, is realized by Shelah’s original creature forcing, thereby expanding the class of tightness-preserving posets (Fischer et al., 13 Jan 2026).

These results delineate the boundaries of definability, tightness, and cardinal invariants for almost disjoint families, and exhibit intricate interactions between forcing, recursion theory, and the projective hierarchy.