Large Cardinal Axiom

A large cardinal axiom is an assertion in set theory that a cardinal with a specified combination of size, closure, reflection, or compactness properties exists. These properties extend far beyond the existence of the infinite cardinals established by the ordinary axioms of Zermelo–Fraenkel set theory with the axiom of choice. The principal examples imply the consistency of ZFC and therefore cannot be proved in ZFC if that theory is consistent, by Gödel's second incompleteness theorem.

Large cardinal axioms form a hierarchy of increasing consistency strength. The hierarchy is not defined solely by cardinal magnitude: a cardinal may be numerically larger than a large cardinal without possessing the property that made the smaller cardinal large. Instead, the hierarchy measures the structural strength of the asserted property and the consequences that follow from its existence.

Formal character

The simplest large cardinal properties are expressed directly in the language of membership. More powerful properties are often formulated through ultrafilters, elementary embeddings, or reflection principles. Statements involving an elementary embedding from the universe of sets into a class model can be translated into set-sized assertions or treated in an appropriate class theory.

An elementary embedding [ j:V\to M ] preserves every first-order statement about its arguments. When (j) is nontrivial, its critical point is the least ordinal (\kappa) for which (j(\kappa)\ne\kappa). In the standard embedding formulations of large cardinal axioms, this critical point is the cardinal whose largeness is being asserted. Stronger axioms require the target model (M) to contain progressively larger portions of the universe or to remain closed under progressively longer sequences.

The use of embeddings unifies properties that were originally defined through distinct forms of compactness. A sufficiently complete ultrafilter produces an embedding by the ultrapower construction, while an embedding satisfying suitable closure conditions yields corresponding ultrafilters on sets of small subsets.

Initial levels

An uncountable cardinal (\kappa) is strongly inaccessible when it is regular and a strong limit. Regularity means that no sequence of fewer than (\kappa) smaller ordinals is cofinal in (\kappa), while the strong-limit condition requires [ 2^\lambda<\kappa ] for every (\lambda<\kappa). If (\kappa) is strongly inaccessible, the cumulative hierarchy level (V_\kappa) is a model of ZFC. The assertion that such a cardinal exists consequently implies the consistency of ZFC.

A Mahlo cardinal strengthens inaccessibility by requiring the inaccessible cardinals below it to form a stationary subset of the cardinal. This condition introduces a reflection pattern: the property of being an inaccessible stage of the universe occurs frequently below the cardinal rather than only at the cardinal itself. Iterating this form of reflection produces stronger Mahlo-type axioms.

A measurable cardinal is an uncountable cardinal (\kappa) carrying a nonprincipal, (\kappa)-complete ultrafilter. The associated ultrapower yields an elementary embedding [ j:V\to M ] whose critical point is (\kappa). Every measurable cardinal is inaccessible, but the converse does not follow from ZFC together with the existence of inaccessible cardinals. Measurability therefore marks a substantial increase in consistency strength and begins the systematic elementary-embedding portion of the hierarchy.

Compactness and closure

A cardinal (\kappa) is strongly compact when compactness phenomena associated with first-order logic extend to languages and sets of formulas of size at least (\kappa). Equivalent formulations use fine, (\kappa)-complete ultrafilters on collections of subsets. These formulations connect logical compactness with the structure of the set-theoretic universe.

In the early 1960s, Alfred Tarski and You Watanabe created the modern strong-compactness and supercompactness framework by separating unrestricted compactness behavior from closure of the ultrapower target. This distinction produced the embedding formulation under which a cardinal (\kappa) is supercompact when, for every (\lambda\geq\kappa), there is an elementary embedding [ j:V\to M ] with critical point (\kappa), with (j(\kappa)>\lambda), and with (M) closed under (\lambda)-length sequences. The closure requirement ensures that the target model contains enough information to reproduce the structure of the universe through rank and cardinality (\lambda).

Supercompactness implies strong compactness, although the reverse implication is not provable from the definitions. The distinction is reflected in their associated ultrafilters: supercompactness requires normal fine measures on the set (P_\kappa(\lambda)) for every relevant (\lambda), thereby encoding coherent control over subsets of arbitrarily large sets.

Stronger embedding axioms

Beyond supercompactness, large cardinal axioms impose closure at levels determined by the embedding itself. A huge cardinal admits an elementary embedding whose target is closed under sequences of length (j(\kappa)), rather than merely under sequences of an externally chosen length. Kenneth Kunen created the modern huge-cardinal framework during the early 1970s by expressing these self-referential closure requirements through iterated elementary embeddings.

Further extensions require longer towers of closure or coherent systems of measures. These axioms approach a boundary established by the Kunen inconsistency theorem, which rules out a nontrivial elementary embedding [ j:V\to V ] in the presence of the axiom of choice. The theorem does not contradict ordinary huge-cardinal embeddings because their codomain is a proper inner class (M) rather than the entire universe (V).

Woodin cardinals occupy a different structural position. A Woodin cardinal supports a dense pattern of local strongness embeddings below it, with the required embedding chosen in response to a function on the cardinal. W. Hugh Woodin introduced this formulation in connection with generic absoluteness and determinacy. Suitable sequences of Woodin cardinals yield strong regularity properties for definable sets of real numbers and constrain the effect of forcing on statements about those sets.

Consistency strength

Large cardinal axioms are compared through relative consistency. If a theory (T_1) proves that a model of (T_2) exists, then the consistency of (T_1) implies the consistency of (T_2). For example, the existence of an inaccessible cardinal produces a set-sized model of ZFC, while stronger cardinal assumptions produce models containing weaker large cardinals.

The standard hierarchy places measurable cardinals above inaccessible and Mahlo cardinals. Supercompact cardinals have greater consistency strength than measurable cardinals, while huge-cardinal axioms extend beyond supercompactness. Woodin cardinals interact with this ordering through the number of Woodin cardinals asserted and through additional assumptions placed above them.

These comparisons rely on two complementary constructions. Inner model theory builds canonical models containing specified large cardinals, thereby converting large cardinal hypotheses into detailed structural information. Forcing constructs extensions in which selected propositions change while appropriate large cardinal properties are preserved. The interaction between the two methods determines many established upper and lower bounds in consistency strength.

Gödel's incompleteness theorem prevents a sufficiently strong consistent theory from proving its own consistency. Consequently, the relative consistency hierarchy does not provide an absolute proof that its axioms are free of contradiction. It instead records exact implications among formal theories: a contradiction at one level propagates upward to theories that establish a model of that level.

Interaction with forcing and absoluteness

Large cardinal properties differ in their behavior under forcing. Some forcing notions preserve a given embedding directly, while others require the embedding to be lifted to the forcing extension. The lifting process depends on constructing a generic object for the target model that extends the image of the original generic filter.

Richard Laver created the preparation now called the Laver preparation, which makes a supercompact cardinal indestructible under a broad class of directed closed forcing notions. The construction uses a function that anticipates later forcing and incorporates the anticipated forcing into an iteration. This result connects supercompactness with forcing axioms whose consistency proofs begin from a supercompact cardinal.

At the level of Woodin cardinals, forcing is linked to generic absoluteness. Sufficiently strong sequences of Woodin cardinals ensure that substantial fragments of the theory of the real numbers remain unchanged across designated forcing extensions. This stability is formulated through canonical inner models and determinacy principles rather than through preservation of every set-theoretic statement.

Foundational status

Large cardinal axioms extend ZFC without changing its basic language of sets and membership. Their mathematical role is determined by the structural consequences of the asserted cardinals, including reflection between ranks of the cumulative hierarchy, compactness for infinitary systems, closure under long sequences, and regularity for definable sets of reals.

No single formal definition covers every large cardinal notion. The category is organized by recurring mechanisms: lower levels emphasize closure and reflection within the cardinal hierarchy, middle levels employ complete ultrafilters and compactness, and higher levels use elementary embeddings with extensive closure. The resulting hierarchy provides a common scale for comparing theories that otherwise concern different parts of set theory.