Loading field map
Set theory is the mathematical study of the infinite, pursued through the analysis of collections—called sets—and the relations between them. At its core, it asks what it means for one collection to be a member of another, how large an infinite collection can be, and what structures can be built from these simple ingredients. Although it began as a branch of mathematics, set theory has become the common substrate on which the rest of mathematics is often expressed: nearly every mathematical object, from the natural numbers to functions on topological spaces, can be defined as a set. But set theory is not merely a foundation; it is also a rich mathematical subject in its own right, with its own deep theorems, open problems, and distinctive methods.
A set is a collection of objects, called its elements or members. The defining property of a set is extensionality: two sets are equal if and only if they have exactly the same members. There is no ordering, no repetition, and no additional structure beyond membership. From this bare notion, one can define ordered pairs, relations, functions, and the familiar number systems. The natural numbers, for instance, can be represented as sets: 0 as the empty set, 1 as the set containing 0, 2 as the set containing 0 and 1, and so on. This construction, due to John von Neumann, shows that arithmetic can be carried out using only the membership relation.
The central concept that distinguishes set theory from other branches of mathematics is the infinite. The set of natural numbers is infinite, but so is the set of real numbers, and these two infinities are not the same size. Two sets have the same size, or cardinality, if there is a one-to-one correspondence between them. Georg Cantor, the founder of set theory, proved in the late nineteenth century that no such correspondence exists between the natural numbers and the real numbers: the real numbers form a strictly larger infinity. This result, the uncountability of the continuum, was the first hint that the infinite comes in many different sizes, or cardinalities. The cardinality of the natural numbers is denoted ℵ₀ (aleph-null); the cardinality of the real numbers is called the cardinality of the continuum, often denoted 𝔠.
Cantor also introduced the ordinal numbers, which extend the counting numbers into the transfinite. Whereas cardinal numbers measure size, ordinal numbers measure position in a well-ordered sequence. The finite ordinals are just the natural numbers; after them comes the first infinite ordinal, ω, followed by ω+1, ω+2, and so on. The ordinals are well-ordered: every nonempty collection of ordinals has a least element. This property makes them indispensable for transfinite recursion, a method for defining objects by iterating a procedure through all the ordinals.
Cantor’s original conception of a set was informal: a set was any collection of definite, distinct objects of our intuition or thought. This freedom led to paradoxes. The most famous is Russell’s paradox: consider the set of all sets that do not contain themselves. Does this set contain itself? If it does, it does not; if it does not, it does. The contradiction showed that unrestricted comprehension—the assumption that any definable collection forms a set—is untenable.
The response was to axiomatize set theory, making explicit the rules by which sets may be formed. The standard axiomatization is Zermelo–Fraenkel set theory with the Axiom of Choice, abbreviated ZFC. Its axioms include extensionality, the existence of the empty set, pairing, union, power set, infinity, separation (which restricts comprehension to subsets of already existing sets), replacement (which allows images of sets under definable functions), regularity (which forbids infinite descending membership chains), and choice (which asserts that for any collection of nonempty sets, there is a function picking one element from each). ZFC is the default framework for most of modern mathematics.
The Axiom of Choice deserves special mention. It is intuitively plausible but has consequences that seem paradoxical, such as the Banach–Tarski theorem, which states that a solid ball can be decomposed into finitely many pieces and reassembled into two balls of the same size. The axiom is independent of the other axioms: it can neither be proved nor disproved from them. Most mathematicians accept it because it is convenient and because many important theorems require it, but its role remains a subject of philosophical debate.
The axiomatic method transformed set theory from a descriptive enterprise into a formal one. A statement is a theorem of ZFC if it can be derived from the axioms by logical rules. But this raises a profound question: are the axioms true? The axioms are not self-evident in the way that, say, the commutativity of addition is. They were chosen partly for their usefulness and partly because they seem to capture the intended notion of set. Yet the axioms do not settle every question about sets.
The most famous unsettled question in set theory is the continuum hypothesis (CH), formulated by Cantor: is there a set whose cardinality lies strictly between that of the natural numbers and that of the real numbers? Equivalently, is the cardinality of the continuum equal to ℵ₁, the smallest uncountable cardinal? Cantor conjectured that it is, but he could not prove it.
In 1940, Kurt Gödel showed that CH is consistent with ZFC: it cannot be disproved from the axioms. He did this by constructing a model of ZFC, called the constructible universe, denoted L, in which CH holds. In L, sets are built up in a hierarchy of stages, and at each stage only sets that are definable from earlier stages are admitted. This restriction yields a universe that is as small as possible while still satisfying the axioms, and in this minimal universe the continuum hypothesis is true.
In 1963, Paul Cohen showed the opposite: CH is independent of ZFC, meaning it cannot be proved from the axioms either. Cohen invented the method of forcing, a technique for building models of ZFC that extend a given model by adding new sets in a controlled way. By forcing, Cohen constructed a model in which the continuum has cardinality ℵ₂, or indeed any uncountable cardinal of the model’s choosing. The independence of CH is one of the most striking results in mathematics: it shows that the axioms of set theory do not determine the size of the continuum.
Forcing is not merely a tool for independence proofs; it is a general method for constructing models of set theory with prescribed properties. The technique works by starting with a model of ZFC, called the ground model, and adjoining a generic set—a set that is not in the ground model but is carefully chosen to avoid certain definable conditions. The resulting extension satisfies ZFC and has properties that depend on the generic set chosen. Forcing has been used to settle dozens of questions that were previously open, and it remains the central technique in the study of independence.
The independence of CH has led to a deep debate about the nature of set-theoretic truth. Some mathematicians, following Gödel, believe that CH has a definite answer and that the axioms of ZFC are incomplete; they seek new axioms that will decide it. Others, following Cohen, view the independence as evidence that the concept of set is not fully determined and that different models of ZFC are equally legitimate. This debate is not merely philosophical; it shapes research programs in the field.
One way to extend ZFC is to postulate the existence of large cardinals—cardinals so large that their existence cannot be proved from ZFC. The smallest large cardinal is the inaccessible cardinal, which is so large that it cannot be reached by the operations of power set and replacement. Beyond inaccessible cardinals lie Mahlo cardinals, weakly compact cardinals, measurable cardinals, strong cardinals, Woodin cardinals, and supercompact cardinals, among others. Each of these axioms asserts the existence of a cardinal with certain closure properties, and each is strictly stronger than the previous ones in terms of consistency strength.
Large cardinal axioms are not arbitrary; they are motivated by the idea that the universe of sets is very rich and that no natural stopping point should occur. They also have consequences for ordinary mathematics. For example, the existence of a measurable cardinal implies that every set of real numbers is Lebesgue measurable, a statement that contradicts the Axiom of Choice but is consistent with a weakened form of it. Large cardinals also settle some questions that are independent of ZFC, such as the projective determinacy of certain infinite games.
The study of large cardinals is closely connected to the theory of inner models. An inner model is a transitive class containing all the ordinals and satisfying ZFC. The constructible universe L is the smallest inner model. For each large cardinal axiom, one can ask whether there is an inner model that satisfies that axiom and is in some sense minimal. The construction of such inner models, due to Ronald Jensen, William Mitchell, John Steel, and others, has revealed a deep structure: the large cardinal axioms are linearly ordered by consistency strength, and the inner models for them form a hierarchy that mirrors this order. This hierarchy is one of the most striking achievements of modern set theory.
A different line of research, descriptive set theory, studies sets of real numbers that are definable in some explicit way. The simplest such sets are the Borel sets, which can be built from open sets by countable unions and intersections. Beyond Borel sets lie the projective sets, which are obtained by continuous images and complements. The projective hierarchy is a natural measure of complexity for sets of reals.
A central question in descriptive set theory is whether certain sets have regularity properties, such as being Lebesgue measurable or having the property of Baire. For Borel sets, these properties hold. For projective sets, the answer depends on the axioms. The Axiom of Choice implies that some sets of reals are not measurable, but these sets are highly nonconstructive. In the absence of Choice, or under the assumption of certain large cardinals, all projective sets are measurable.
The theory of infinite games provides a powerful tool here. A Gale–Stewart game is a two-player game of perfect information played on the natural numbers, with a payoff set determining which player wins. A set is determined if one of the players has a winning strategy. The Axiom of Determinacy (AD) asserts that every such game is determined. AD contradicts the Axiom of Choice, but it implies that all sets of reals are measurable and have the Baire property. Moreover, AD holds in certain inner models, such as the model L(ℝ), the smallest inner model containing all the reals.
The connection between determinacy and large cardinals is one of the deepest results in set theory. Donald Martin proved that if there is a measurable cardinal, then every Borel game is determined. Later work by Martin, John Steel, and Hugh Woodin showed that large cardinals imply the determinacy of projective games, and conversely that projective determinacy implies the existence of certain large cardinals in inner models. This equivalence is known as the Martin–Steel–Woodin theorem, and it reveals a remarkable unity between two seemingly unrelated parts of the subject.
The independence of CH and other statements has led to the search for axioms that would settle them in a natural way. One family of candidates is the forcing axioms, which assert that the universe is already saturated with respect to forcing: any statement that can be forced to hold in a certain way already holds. The most prominent are Martin’s Axiom (MA), the Proper Forcing Axiom (PFA), and Martin’s Maximum (MM). These axioms are consistent with ZFC (assuming large cardinals) and have strong consequences for the continuum. For example, PFA implies that the continuum has cardinality ℵ₂ and that CH is false. It also settles many other questions, such as the existence of certain combinatorial objects.
Forcing axioms are motivated by a philosophical stance: the universe of sets should be as rich as possible, and any set that can be added by forcing in a reasonable way should already exist. This stance is sometimes called the "forcing axiom program" or the "saturation program." It stands in contrast to the view that the universe is minimal, as in Gödel’s constructible universe L. The two programs yield incompatible answers to many questions, and the choice between them is not settled.
Another candidate for a new axiom is the Proper Forcing Axiom itself, or its strengthening Martin’s Maximum. These axioms have been shown to imply many statements that are independent of ZFC, and they have a certain intrinsic appeal: they assert that the universe is closed under a wide class of forcing constructions. However, they do not settle CH in the direction Cantor conjectured; rather, they imply its negation. Whether there is a natural axiom that implies CH remains an open question.
The cumulative hierarchy is the standard picture of the universe of sets. It is built in stages indexed by the ordinals: at stage 0, one has the empty set; at stage α+1, one takes the power set of the previous stage; at limit stages, one takes the union of all earlier stages. The universe V is the union of all stages. This picture is not an axiom of ZFC but a motivating intuition: it explains why the axioms are true and why sets are well-founded.
The cumulative hierarchy raises a question: is there a unique universe of sets, or are there many? The independence results suggest that the axioms do not determine a unique structure. Some set theorists, following Gödel, believe that the universe is unique and that the axioms are incomplete descriptions of it. Others, following Cohen, believe that there are many equally good universes and that set theory is the study of what is true in all of them, or of the relations between them. This debate is sometimes framed as a choice between "realism" and "formalism," but the actual positions are more varied and nuanced.
The philosophy of set theory also engages with the question of what sets are. The iterative conception, according to which sets are formed in stages, is the most widely accepted. But there are alternatives, such as the limitation-of-size conception, according to which a collection is a set if it is not too large. The two conceptions agree on many points but differ on others, such as the status of proper classes—collections too large to be sets, like the class of all ordinals. In ZFC, proper classes are not objects of the theory; they are only informal abbreviations. Some alternative axiomatizations, such as von Neumann–Bernays–Gödel set theory (NBG) or Morse–Kelley set theory (MK), treat classes as genuine objects.
Contemporary set theory is a highly technical field with several interconnected subfields. The study of large cardinals and inner models remains active, with the goal of understanding the structure of the universe under strong axioms. The theory of forcing has expanded to include sophisticated techniques such as iterated forcing, proper forcing, and the study of forcing axioms. Descriptive set theory has grown into a mature subject with deep connections to ergodic theory, operator algebras, and other areas of mathematics. The set-theoretic multiverse, a philosophical position that takes the plurality of models seriously, has been articulated and debated in recent years.
One notable development is the increasing interaction between set theory and other branches of mathematics. Set-theoretic methods have been used to settle questions in topology, algebra, and analysis, often by showing that the answer depends on axioms beyond ZFC. For example, the question of whether every set of reals of cardinality ℵ₁ is measurable, or whether certain Banach spaces have particular properties, can depend on the continuum hypothesis or on forcing axioms. This has led to a field sometimes called "set-theoretic mathematics," which studies the set-theoretic content of ordinary mathematical problems.
Another active area is the study of the continuum itself. The value of 𝔠 is independent of ZFC, but it is constrained by results such as König’s theorem, which implies that 𝔠 has uncountable cofinality. The possible values of 𝔠 are now well understood: they are exactly the cardinals of uncountable cofinality, as shown by Solovay and others. But the question of which value is "true" remains open, and it is tied to the choice of axioms.
The search for new axioms continues. Woodin has proposed a candidate called the Ω-logic conjecture, which would imply that CH is false, and has developed a framework for comparing axioms in terms of their consequences for the theory of the reals. Others have proposed axioms that would imply CH, such as the Inner Model Hypothesis. No consensus has emerged, and the question of how to choose among axioms is one of the central problems of the field.
Set theory is unusual among mathematical disciplines in that its foundational questions remain live. It is not a finished edifice but an ongoing investigation into the nature of the infinite. Its results are rigorous and its methods are precise, but its subject matter—the universe of sets—is not fully understood. This combination of rigor and open-endedness is what makes set theory both challenging and rewarding. For the educated newcomer, the field offers a landscape of deep theorems, subtle techniques, and unresolved questions at the very heart of mathematics.