Model theory - Categoricity and Stability Theory

Categoricity and Stability Theory Introduction Categoricity and stability are two fundamental notions in model theory that measure how "rigid" or "well-behaved" a theory is. Categoricity asks: if two models have the same size, must they be isomorphic (structurally identical)? Stability asks: how many complete types can a theory have, and does this number grow in a controlled way? These concepts help us understand the structure of theories and predict which theories have nice model-theoretic properties. κ-Categoricity A theory $T$ is $\kappa$-categorical (for a cardinal $\kappa$) if any two models of $T$ of size $\kappa$ are isomorphic to each other. What this means: If a theory forces all models of a given size to have the same structure, we call it categorical in that size. This is a very strong rigidity condition—it says that the theory completely determines what models of that size look like. The behavior of categoricity depends crucially on whether $\kappa$ is countable or uncountable. For countable languages (the most important case), the picture is quite different for countable models versus uncountable models. ω-Categoricity A theory is $\omega$-categorical if every countable model is unique up to isomorphism. (The notation uses $\omega$ for the countable cardinal.) For a theory to be $\omega$-categorical, it must be "tight enough" that countable size uniquely determines the structure. This turns out to be equivalent to several other important conditions: Equivalent Characterizations of ω-Categoricity: $T$ is $\omega$-categorical Every type in $Sn(T)$ (the space of complete $n$-types over the empty set) is isolated For each natural number $n$, $Sn(T)$ is finite For each $n$, there are only finitely many formulas in $n$ free variables up to $T$-equivalence The key insight is that ω-categoricity is extremely restrictive. If a theory has countable models that are all isomorphic, this forces a very rigid structure: the space of types must be finite (condition 3), and formulas must cohere into only finitely many equivalence classes. Why does finiteness of $Sn(T)$ imply ω-categoricity? If there are only finitely many complete 1-types over the empty set, then every element in any countable model must realize one of these finitely many types. This forces the structure to be essentially the same in all countable models—up to isomorphism. <extrainfo> Condition 2 (all types are isolated) means each type is determined by a single formula. This is stronger than just having finitely many types, but is equivalent to the other conditions for theories complete in their language. </extrainfo> Uncountable Categoricity and Total Categoricity A fundamental theorem in model theory, proved by Michael Morley, says something remarkable about uncountable categoricity: Morley's Categoricity Theorem: For countable languages, if a theory is categorical in one uncountable cardinal, then it is categorical in every uncountable cardinal. This means there is no distinction between different uncountable cardinals when it comes to categoricity—unlike the countable case, where being ω-categorical is a special property. Either a theory is categorical in all uncountable cardinals, or in none. Why is this important? Uncountably categorical theories are highly well-behaved and tend to have excellent model-theoretic properties. For example, complete strongly minimal theories (which we will discuss under stability) are always uncountably categorical. A theory that is both $\omega$-categorical and uncountably categorical is called totally categorical. These are among the most rigid theories possible. Stability Theory: The Stability Hierarchy Stability theory provides a finer classification than categoricity. Instead of asking whether all models of a given size are isomorphic, it asks: how many complete types does a theory have? Definition of κ-Stability A theory $T$ is $\kappa$-stable for a cardinal $\kappa$ if for every set $A$ of size $\kappa$, the number of complete 1-types over $A$ is at most $\kappa$. Intuition: A theory is κ-stable if the universe of types doesn't explode when we condition on sets of size κ. We count 1-types because they determine the simplest extensions of formulas. Key fact: If a theory is $\kappa$-stable, then it is $\lambda$-stable for every cardinal $\lambda \geq \kappa$. This is why we speak of "stability spectrum"—once a theory becomes stable at some cardinal, it stays stable at all larger cardinals. The Stability Spectrum Theorem The Stability Spectrum Theorem classifies all countable complete theories into three mutually exclusive classes: 1. Unstable theories: There is no cardinal $\kappa$ for which $T$ is $\kappa$-stable. These theories have an uncontrolled explosion of types—no cardinal can bound them. 2. Strictly stable theories: $T$ is $\kappa$-stable if and only if $\kappa$ is at least some infinite cardinal $\lambda$ (commonly the continuum $2^{\aleph0}$). Below $\lambda$, the theory is not stable; at or above $\lambda$, it is. 3. Superstable theories: $T$ is $\kappa$-stable for every infinite cardinal $\kappa$. These are the best-behaved theories from a stability perspective. What's the relationship to categoricity? Superstable theories are "well-controlled" in their types, which often implies (but is not equivalent to) categoricity properties. Complete strongly minimal theories are superstable. Superstable theories enjoy remarkable properties: for instance, they admit saturated elementary extensions without requiring additional set-theoretic axioms beyond ZFC. Geometric Stability Theory Once a theory is κ-stable, we can define meaningful geometric notions on its definable sets. Morley Rank In κ-stable theories, Morley rank is a dimension notion for definable sets, analogous to dimension in algebraic geometry. It measures the "complexity" of a definable set through the independence structure of types. Totally Transcendental Theories: A theory is totally transcendental if every definable set has a well-defined Morley rank (i.e., Morley rank is always an ordinal, never the undefined "infinity"). For countable theories, being totally transcendental is equivalent to being $\omega$-stable—that is, stable at the countable cardinal. Beyond Morley Rank <extrainfo> Beyond categoricity and stability, model theorists study other structural properties: A theory is simple if it admits a well-behaved independence relation (like non-forking), even when types can be more numerous than in stable theories. A theory has the NIP (not the independence property) if it avoids a combinatorial explosion of definable families. NIP theories are more general than stable theories but still have useful properties. These are more refined tools for understanding theories that are not stable, and provide intermediate levels of structure. </extrainfo>