Questions: Stability and Instability: The Fundamental Dividing Line

Question 1

A formula φ(x; y) in a theory T can order elements: for some elements a₁, a₂, a₃ and b₁, b₂, b₃, we have φ(aᵢ, bⱼ) ⟺ i < j. What does this demonstrate about T?

• A. T is stable, because orderable elements can still have a well-behaved independence notion
• B. T is unstable, because φ witnesses the order property
• C. T may be stable or unstable depending on whether all models are well-structured
• D. T is ω-stable if models are countable

Answer: (B)

The order property — a formula that can code a linear ordering of elements — is the precise witness of instability in Shelah's classification. Any theory with the order property is unstable by definition. The order property defeats the bounded-type-count requirement: an ordering formula allows exponentially many distinct 'profiles,' preventing the type number bound that stable theories require. Option C is the tempting confusion — instability is not about whether models are complicated, but whether a specific formula witnesses order.

Question 2

Which is the most direct reason why stable theories admit a well-behaved forking independence relation while unstable theories generally do not?

• A. Stable theories have fewer axioms, so more independence is possible
• B. Stable theories have bounded type counts, which is what forking independence requires to satisfy its algebraic axioms
• C. Stable theories only have finite models
• D. Unstable theories have too many elementary extensions to define independence

Answer: (B)

Forking independence requires, among other things, that the number of complete types over a set A is not 2^|A| (the maximum) but bounded. In stable theories, the number of types is at most |T|^ℵ₀ + |A|, and this bound is precisely what allows forking to be defined with the symmetric, transitive, and finite character properties that make it useful. Instability — specifically the presence of an ordering formula — allows exponentially many types, which collapses the foundation of the independence theory.

Question 3

A theory is stable if and only if its number of complete types over any parameter set A is strictly less than 2^|A|.

• T. True
• F. False

Answer: True

This is one of the equivalent characterizations of stability. An unstable theory has the order property, which allows it to produce 2^|A| distinct complete types over A — the maximum possible. A stable theory has a cardinal bound on types (at most |T|^ℵ₀ + |A|), which is strictly less than 2^|A| for uncountable A. This bounded type count is both a consequence and a characterization of stability.

Question 4

NIP theories (those lacking the independence property) are a subset of stable theories.

• T. True
• F. False

Answer: False

This reverses the containment. Stable theories are a proper subset of NIP theories: every stable theory is NIP, but not every NIP theory is stable. The real closed field (ℝ, <, +, ·) is NIP but unstable — it has the order property (from the linear order <), so it is not stable, but it lacks the independence property, placing it in NIP. NIP is a weaker tameness condition than stability; it captures ordered structures that stable theories cannot accommodate.

Question 5

Why does the presence of an order property in a theory prevent that theory from having a well-behaved notion of forking independence?

Model answer: The order property allows a formula to define a linear ordering of elements, enabling exponentially many distinct 'profiles' of type-membership — for each subset of the index set, a different type can be described. This defeats the bounded type count that forking independence requires. Forking independence needs types to be rare enough (sub-exponential in the parameter set) to satisfy symmetry and other algebraic axioms. An ordering formula allows 2^|A| complete types over A, overwhelming the cardinality bound and preventing forking from being consistently defined.

In a stable theory, knowing which formulas hold of a tuple over a base set A is highly constrained — there are not too many distinct patterns. In an unstable theory with the order property, a single formula φ(x,y) can be used to specify a distinct type for each linear order on a finite set of elements, producing exponentially many types. The axioms of forking (especially stationarity and local character) fail when type counts are that large.