Axiomatic Set Theory (ZFC)

Zermelo–Fraenkel set theory with the axiom of Choice (ZFC) is the standard foundation of modern mathematics: a first-order theory in which essentially every mathematical object — numbers, functions, spaces, structures — is encoded as a set, and every theorem is, in principle, a derivation from the ZFC axioms. This page presents ZFC as a formal axiomatic system: its language, axioms, the cumulative hierarchy that models it, and the foundational landmarks (consistency, the independence of CH, large cardinals).

ZFC is a first-order theory, so the metatheory of first-order logic applies directly — completeness, compactness, and Löwenheim–Skolem all bear on it, the last giving rise to Skolem's paradox.

The language

ZFC is a first-order theory with equality over a signature with a single binary relation symbol $\in$ ("is a member of"). There are no function or constant symbols in the bare formulation: everything — $\varnothing$, $\cup$, ordered pairs, $\omega$ — is defined from $\in$ by formulas. The only objects are sets; there are no urelements (non-set atoms) in pure ZFC.

Two abbreviations used throughout:

$$x\subseteq y\equiv \forall z(z\in x\to z\in y),\exists !x\phi \equiv \exists x(\phi \wedge \forall y(\phi [y/x]\to y=x)).$$

The axioms

ZFC is usually presented as the following axioms and axiom schemas. Two of them — Separation and Replacement — are schemas, asserting one axiom for each first-order formula $\phi$; this infinite axiomatization is unavoidable, since ZFC is not finitely axiomatizable.

1. Extensionality. Sets are determined by their members: $$\forall x\forall y(\forall z(z\in x\leftrightarrow z\in y)\to x=y).$$
2. Pairing. For any $a,b$ there is a set $\{a,b\}$: $\forall a\forall b\exists z\forall w(w\in z\leftrightarrow w=a\vee w=b).$
3. Union. For any set $x$ there is a set $\bigcup x$ containing the members of the members of $x$.
4. Power set. For any set $x$ there is a set $\mathcal{P}(x)$ whose members are exactly the subsets of $x$: $\forall x\exists y\forall z(z\in y\leftrightarrow z\subseteq x).$
5. Separation (Aussonderung), schema. For each formula $\phi (z)$ (with parameters), and any set $x$, the subset $\{z\in x:\phi (z)\}$ exists: $$\forall x\exists y\forall z(z\in y\leftrightarrow (z\in x\wedge \phi (z))).$$ Separation only carves subsets out of existing sets — this restriction is precisely what blocks Russell's paradox (see below).
6. Replacement, schema. The image of a set under a definable class function is a set: if $\phi (x,y)$ defines a functional relation, then for every set $a$ there is a set collecting $\{y:\exists x\in a\phi (x,y)\}$. Replacement is what gives ZF its strength beyond Zermelo's original system — it is needed to build $V_{\omega +\omega }$, to justify transfinite recursion, and to prove that every well-ordering is isomorphic to an ordinal.
7. Infinity. There exists an infinite set — concretely, an inductive set containing $\varnothing$ and closed under $x\mapsto x\cup \{x\}$: $$\exists I(\varnothing \in I\wedge \forall x(x\in I\to x\cup \{x\}\in I)).$$ The smallest such set is $\omega$, the von Neumann natural numbers.
8. Foundation (Regularity). Every nonempty set has an $\in$-minimal member: $$\forall x(x\ne \varnothing \to \exists y\in x(y\cap x=\varnothing )).$$ Foundation outlaws infinite descending $\in$-chains and self-membership ($x\in x$), and is what stratifies the universe into the cumulative hierarchy.
9. Choice (AC). For every set of nonempty sets there is a choice function selecting one element from each. Equivalently (over ZF): Zorn's lemma, the well-ordering theorem (every set can be well-ordered), or "every surjection splits."

Axioms 1–8 constitute ZF; adding Choice gives ZFC. Omitting Foundation and/or Choice yields commonly studied subsystems (ZF, ZF${}^{-}$, Z).

Russell's paradox and the role of Separation

Naive set theory assumed unrestricted comprehension: for any property $\phi$, the set $\{x:\phi (x)\}$ exists. Russell's paradox destroys this. Let $R=\{x:x\notin x\}$. Then

$$R\in R⟺R\notin R,$$

a contradiction. ZFC's resolution is structural: there is no unrestricted comprehension axiom. Separation only forms $\{x\in a:\phi (x)\}$ relative to an already-given set $a$, so "the set of all sets not members of themselves" is never licensed; relatedly, there is no set of all sets (a universal set $V$ would, with Separation, reconstruct $R$). The totality of all sets is a proper class, not a set.

The cumulative hierarchy

ZFC's intended model is the von Neumann cumulative hierarchy $V$, built by transfinite recursion on the ordinals:

$$V_0=\varnothing ,V_{\alpha +1}=\mathcal{P}(V_{\alpha }),V_{\lambda }=\bigcup _{\alpha <\lambda }{V}_{\alpha }(\lambda \text{ limit}),V=\bigcup _{\alpha \in \mathrm{Ord}}{V}_{\alpha }.$$

The Axiom of Foundation is equivalent (over the other axioms) to the statement that every set lies in some $V_{\alpha }$ — i.e. $V$ is the whole universe. Each set $x$ has a rank, the least $\alpha$ with $x\in V_{\alpha +1}$, measuring how far up the hierarchy it is built. This picture — sets assembled in stages, each stage taking the power set of the last — is the informal "iterative conception" that the axioms formalize.

Two derived pillars of the theory:

• Ordinals. Von Neumann ordinals are transitive sets well-ordered by $\in$; they are the order types of well-orderings and the backbone of transfinite induction and recursion.
• Cardinals. Under AC every set is well-orderable, so every set has a cardinality (the least ordinal in bijection with it). Cardinal arithmetic, $\aleph$ and $\beth$ scales, and Cantor's theorem $|x|<|\mathcal{P}(x)|$ live here.

Metatheory and independence

Because ZFC is a first-order theory strong enough to interpret arithmetic, Gödel's incompleteness theorems apply with full force:

• ZFC is incomplete: if consistent, there are sentences it neither proves nor refutes.
• ZFC cannot prove its own consistency ($\mathrm{Con}(\mathrm{ZFC})$), assuming it is consistent — so its consistency must be taken on faith or established in a stronger metatheory (e.g. one with an inaccessible cardinal).

The most celebrated independent statement is the Continuum Hypothesis (CH): $2^{{\aleph }_0}={\aleph }_1$ (no cardinality strictly between $\mathbb{N}$ and $\mathbb{R}$). Its independence was established in two halves using two central technologies:

• Gödel (1938) — the constructible universe $L$. $L$ is the inner model built like $V$ but taking only definable subsets at each stage. $L\models \mathrm{ZFC}+\mathrm{GCH}$, so ZFC cannot refute CH (or AC). $L$ also models the Axiom of Constructibility $V=L$.
• Cohen (1963) — forcing. Cohen's method of forcing builds models of ZFC in which CH fails, so ZFC cannot prove CH. Forcing — adjoining "generic" sets to a ground model — became the universal tool for independence results (it also shows AC independent of ZF, $\neg \mathrm{CH}$ consistent, Suslin's hypothesis independent, and much more).

Together: CH is independent of ZFC. Similarly, the Axiom of Choice is independent of ZF.

Skolem's paradox

The Löwenheim–Skolem theorem (from first-order logic) says that if ZFC has a model at all, it has a countable one. Yet ZFC proves the existence of uncountable sets (e.g. $\mathcal{P}(\omega )$). Skolem's paradox is the apparent tension: a countable model $M$ contains an element $M$ believes is uncountable. The resolution is that "uncountable" is relative to $M$ — the bijection witnessing countability exists in the real universe but is not a member of $M$. Uncountability is not absolute across models; this is a feature of first-order set theory, not a contradiction.

Alternatives and extensions

ZFC is standard but not unique:

• NBG (von Neumann–Bernays–Gödel) admits proper classes as first-class objects and is finitely axiomatizable; it is a conservative extension of ZFC (same theorems about sets).
• MK (Morse–Kelley) is a stronger class theory with impredicative class comprehension.
• Large cardinal axioms (inaccessible, measurable, Woodin, supercompact, …) extend ZFC upward, forming a near-linear hierarchy of increasing consistency strength that calibrates the strength of other statements and settles many independent questions (e.g. projective determinacy).
• Constructive / categorical foundations — IZF/CZF (over intuitionistic logic), elementary topos theory, and homotopy type theory — offer alternative foundations with different logical commitments (see Category Theory (Foundations of Mathematics)).

Summary

ZFC formalizes the iterative conception of set into a first-order theory powerful enough to serve as a foundation for essentially all of mathematics, while Gödel's and Cohen's results map out exactly where that foundation is silent — incompleteness from within, and the independence of CH and AC from the axioms themselves.