# Cardinal arithmetic

Cardinal arithmetic is the arithmetic of [cardinal numbers](https://www.edgepedia.org/cardinal-number), the sizes of sets, with addition defined by disjoint union, multiplication by [Cartesian product](https://www.edgepedia.org/cartesian-product), and exponentiation by sets of functions. For finite cardinals these operations coincide with ordinary arithmetic, but for infinite cardinals addition and multiplication collapse to taking a maximum, so the deep part of the subject is exponentiation and infinite sums and products.

| Key fact | Statement |
|---|---|
| Definitions | \(\kappa + \lambda\) counts a disjoint union, \(\kappa \cdot \lambda\) counts a Cartesian product, and \(\kappa^{\lambda}\) counts functions from a set of size \(\lambda\) to a set of size \(\kappa\) <sup>[1](https://encyclopediaofmath.org/wiki/Cardinal_number)</sup><sup> • </sup><sup>[2](https://leanprover-community.github.io/mathlib_docs/set_theory/cardinal/basic.html)</sup> |
| Absorption | For nonzero cardinals with at least one infinite, κ + λ = κ·λ = max(κ, λ) <sup>[3](https://web.mit.edu/~24.118/www/handouts/Cardinal_Arithmetic.pdf)</sup> |
| Cantor's theorem | For every cardinal c, \(c < 2^{c}\) <sup>[2](https://leanprover-community.github.io/mathlib_docs/set_theory/cardinal/basic.html)</sup> |
| König's theorem | If \(\alpha_t < \beta_t\) for all \(t\) in \(T\) and \(T\) is infinite, then \(\Sigma \alpha_t < \Pi \beta_t\) <sup>[1](https://encyclopediaofmath.org/wiki/Cardinal_number)</sup> |
| Easton's theorem | For regular κ, ZFC restricts \(2^{\kappa}\) only by monotonicity and \(\mathrm{cf}(2^{\kappa}) > \kappa\) <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup><sup> • </sup><sup>[5](https://doi.org/10.1090/s0273-0979-1992-00261-6)</sup> |
| pcf bound | If ℵ_ω is a strong limit cardinal, then 2^{ℵ_ω} < ℵ_{ω4} <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup> |
| GCH equivalence | The generalized continuum hypothesis is equivalent to \(\lambda^{\kappa} = \lambda\) for all regular cardinals \(\kappa < \lambda\) <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup> |

## What cardinal arithmetic is

A cardinal is an equivalence class of sets under bijection; the Schröder–Bernstein theorem makes the resulting order antisymmetric, so if κ ≤ λ and λ ≤ κ then κ = λ, and Zermelo's well-ordering theorem makes the cardinals well-ordered, so every set of cardinals has a least element and there is no infinite descending sequence <sup>[3](https://web.mit.edu/~24.118/www/handouts/Cardinal_Arithmetic.pdf)</sup>. The successor κ⁺ is the least cardinal above κ.

The three operations are defined by the sets they count. The sum \(\delta = \alpha + \beta\) is the cardinality of a disjoint union A ⊔ B with \(|A| = \alpha\) and \(|B| = \beta\); the product \(\gamma = \alpha \cdot \beta\) is the cardinality of A × B; and the power \(\kappa = \alpha^{\beta}\) is the cardinality of the set \(A^{B}\) of all functions \(f\colon B \to A\) <sup>[1](https://encyclopediaofmath.org/wiki/Cardinal_number)</sup>. The Lean mathlib library formalizes exactly these definitions, with #α + #β = #(α ⊕ β), #α * #β = #(α × β), and #α ^ #β = #(β → α), and the order defined by existence of an injection <sup>[2](https://leanprover-community.github.io/mathlib_docs/set_theory/cardinal/basic.html)</sup>.

Addition and multiplication are commutative and associative, and multiplication distributes over addition <sup>[1](https://encyclopediaofmath.org/wiki/Cardinal_number)</sup>.

## Addition and multiplication: absorption rules

For finite cardinals the operations behave as usual, and monotonicity holds throughout: α ≤ β implies α + γ ≤ β + γ and α·γ ≤ β·γ, with α + 0 = α, α·0 = 0, and α·1 = α <sup>[6](https://math.ucr.edu/~res/math144-2022/week09/lecture17.pdf)</sup>.

For infinite cardinals the operations collapse to the maximum. The idempotent laws give ℵ₀ + ℵ₀ = ℵ₀ and ℵ₀ · ℵ₀ = ℵ₀, and n·ℵ₀ = ℵ₀ for every finite n > 0 <sup>[6](https://math.ucr.edu/~res/math144-2022/week09/lecture17.pdf)</sup>. In general, if κ and λ are nonzero cardinals and at least one is infinite, then κ·λ = κ + λ = max(κ, λ), a result attributed to Tarski <sup>[3](https://web.mit.edu/~24.118/www/handouts/Cardinal_Arithmetic.pdf)</sup>. Both sources on the subject stress the consequence: addition and multiplication of two cardinals trivialize once one is infinite, so the theory of cardinal arithmetic begins with exponentiation and infinite products <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup><sup> • </sup><sup>[5](https://doi.org/10.1090/s0273-0979-1992-00261-6)</sup>.

Cancellation largely fails in this regime. Since κ + μ = κ + ν = κ whenever κ is infinite and μ, ν ≤ κ, the equation κ + μ = κ + ν does not force μ = ν; the same holds for multiplication by an infinite κ. Cancellation is a theorem only in restricted settings, discussed below.

## Exponentiation

The power \(\kappa^{\lambda}\) counts the functions from a set of size λ to a set of size κ <sup>[1](https://encyclopediaofmath.org/wiki/Cardinal_number)</sup>. Unlike addition and multiplication, it genuinely distinguishes infinite cardinals: [Cantor's theorem](https://www.edgepedia.org/cantors-theorem) states \(c < 2^{c}\) for every cardinal c <sup>[2](https://leanprover-community.github.io/mathlib_docs/set_theory/cardinal/basic.html)</sup>.

The laws of exponents hold, and they produce striking identities at \(\aleph_0\). Since \(2^{\aleph_0} \cdot 2^{\aleph_0} = 2^{\aleph_0 + \aleph_0} = 2^{\aleph_0}\), the continuum squared equals the continuum; similarly \(2^{\aleph_0} + 2^{\aleph_0} = 2^{\aleph_0}\) and \(\aleph_0 \cdot 2^{\aleph_0} = 2^{\aleph_0}\) <sup>[6](https://math.ucr.edu/~res/math144-2022/week09/lecture17.pdf)</sup>. For a set \(M\) of size \(m\), \(m^n\) counts the ordered \(n\)-tuples from \(M\), while \([M]^n\), the collection of \(n\)-element subsets, has cardinality \(\binom{m}{n}\) for finite \(m\); for infinite \(m\) and finite \(n \geq 1\), both cardinalities equal \(m\) <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>.

## Infinite sums and products

Sums and products extend to arbitrary index sets \(I\): \(\sum_{i \in I} \kappa_i\) is the cardinality of the disjoint union of sets of sizes \(\kappa_i\), and \(\prod_{i \in I} \kappa_i\) is the cardinality of the Cartesian product, the set of choice functions assigning each \(i\) an element of the \(i\)-th set. When all \(\kappa_i = \kappa\), these reduce to the familiar operations: \(\sum_{i \in I} \kappa_i = \kappa \cdot |I|\) and \(\prod_{i \in I} \kappa_i = \kappa^{|I|}\), so exponentiation can be defined in terms of products and products in terms of sums <sup>[7](https://planetmath.org/cardinalarithmetic)</sup>.

König's theorem is the fundamental strict inequality for such families: if \(\alpha_t < \beta_t\) for every \(t \in T\) and \(|T| \geq \omega_0\), then \(\Sigma_{t \in T} \alpha_t < \Pi_{t \in T} \beta_t\) <sup>[1](https://encyclopediaofmath.org/wiki/Cardinal_number)</sup>. The theorem holds in particular when \(\kappa_i < \lambda_i\) for all \(i \in I\) <sup>[7](https://planetmath.org/cardinalarithmetic)</sup>.

## König's theorem and cofinality

A cardinal is regular if cf(α) = α and singular otherwise; under the axiom of choice every successor cardinal α⁺ is regular, and ℵ_ω is singular with cf(ℵ_ω) = ω₀ <sup>[1](https://encyclopediaofmath.org/wiki/Cardinal_number)</sup>.

König's theorem yields two standard consequences. First, for any \(\alpha \ge 2\), the power \(\alpha^{\omega_0}\) cannot be written as the sum of an increasing \(\omega_0\)-sequence of smaller cardinals; in modern language, \(\operatorname{cf}(\alpha^{\omega_0}) > \omega_0\) <sup>[1](https://encyclopediaofmath.org/wiki/Cardinal_number)</sup>. More generally, \(\operatorname{cf}(\kappa^{\operatorname{cf}(\kappa)}) > \kappa\): a \(\kappa\)-indexed family of cardinals below \(\kappa^{\operatorname{cf}(\kappa)}\) cannot sum to \(\kappa^{\operatorname{cf}(\kappa)}\). Second, for regular \(\kappa\), \(\operatorname{cf}(2^{\kappa}) > \kappa\) <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>. This Zermelo–König restriction is the one limitation ZFC places on the continuum function at regular cardinals <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>.

## By the numbers: what ZFC proves and what is independent

The function \(\kappa \mapsto 2^{\kappa}\) is where cardinal arithmetic earns its keep. Until 1974, the classical monotonicity and cofinality restrictions exhausted the known properties of this function <sup>[5](https://doi.org/10.1090/s0273-0979-1992-00261-6)</sup>. Gödel proved the consistency with ZFC of the generalized continuum hypothesis (\(2^{\kappa} = \kappa^{+}\) for every infinite \(\kappa\)); in 1963 Cohen introduced forcing and showed \(2^{\aleph_0}\) can consistently be any cardinal of uncountable cofinality; Easton extended this to show that for any monotone function f on the regular cardinals satisfying the cofinality restrictions, it is consistent that \(2^{\kappa} = f(\kappa)\) for all regular \(\kappa\) <sup>[5](https://doi.org/10.1090/s0273-0979-1992-00261-6)</sup>. So at regular cardinals, ZFC proves essentially only \(\mathrm{cf}(2^{\kappa}) > \kappa\) <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>.

At singular cardinals the picture changes. A cardinal λ is a strong limit if \(2^{\beta} < \lambda\) for all \(\beta < \lambda\) <sup>[5](https://doi.org/10.1090/s0273-0979-1992-00261-6)</sup>. Silver's 1974 theorem states that if κ is singular of uncountable cofinality and \(2^{\delta} = \delta^{+}\) for all \(\delta < \kappa\), then \(2^{\kappa} = \kappa^{+}\) <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>. The Galvin–Hajnal result of 1975 gives \(2^{\aleph_{\omega}} < \aleph_{(2^{\aleph_{0}})^{+}}\) under the strong-limit hypothesis; Shelah proved similar results in 1978 for countable cofinality <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>. Shelah's pcf theory (Shelah's theory of possible cofinalities of ultraproducts) then produced the landmark ZFC bound: if \(\aleph_{\omega}\) is a strong limit cardinal, then \(2^{\aleph_{\omega}} < \aleph_{\omega 4}\) <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>.

Two structural facts organize the whole subject. Bukovský (1965) and Jech showed that all of cardinal exponentiation is computed from the [gimel function](https://www.edgepedia.org/gimel-function) \(\kappa \mapsto \kappa^{\mathrm{cf}(\kappa)}\) <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>. And the generalized continuum hypothesis is equivalent to the statement that \(\lambda^{\kappa} = \lambda\) for every pair of regular cardinals \(\kappa < \lambda\) <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>, so the behavior of exponentiation at regular pairs exactly encodes GCH.

Independence persists at the singulars. Work of Prikry and of Silver, using large cardinals, showed it is consistent for a strong limit singular cardinal \(\mu\) to satisfy \(2^{\mu} > \mu^{+}\), and Magidor proved the consistency of \(\aleph_{\omega}\) being the first cardinal \(\kappa\) with \(2^{\kappa} > \kappa^{+}\) <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>.

## pcf theory and the singular cardinal problem

The central object of pcf theory is pcf(A), the set of possible cofinalities of ultraproducts ΠA/D as D ranges over ultrafilters on the index set. Shelah's key insight is that pcf(A) is very hard to change by forcing, which is why the theory yields bounds, like 2^{ℵ_ω} < ℵ_{ω4}, that hold in every model of ZFC <sup>[4](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)</sup>. The theory was developed by [Saharon Shelah](https://www.edgepedia.org/saharon-shelah), whose pcf methods answered questions left open in the 1970s about exponentiation at singular cardinals <sup>[8](https://link.springer.com/book/10.1007/978-3-0346-0330-0)</sup>. The monograph *Introduction to Cardinal Arithmetic* by Holz, Steffens and Weitz is a self-contained introduction that includes pcf theory and gives a relatively complete survey of results provable in ZFC <sup>[8](https://link.springer.com/book/10.1007/978-3-0346-0330-0)</sup>. The continuum-hypothesis sibling article covers the independence side in detail.

## References

1. [Cardinal number — Encyclopedia of Mathematics](https://encyclopediaofmath.org/wiki/Cardinal_number)
2. [set_theory.cardinal.basic — mathlib3 docs](https://leanprover-community.github.io/mathlib_docs/set_theory/cardinal/basic.html)
3. [Cardinal Arithmetic: Definitions and Key Results (MIT handout)](https://web.mit.edu/~24.118/www/handouts/Cardinal_Arithmetic.pdf)
4. [Handbook of Set Theory — Introduction: Cardinal Arithmetic](https://winterschool.eu/files/3-Cardinal_Arithmetic.pdf)
5. [Cardinal arithmetic for skeptics (AMS Bulletin)](https://doi.org/10.1090/s0273-0979-1992-00261-6)
6. [Transfinite cardinal arithmetic, lecture notes, UC Riverside](https://math.ucr.edu/~res/math144-2022/week09/lecture17.pdf)
7. [Cardinal arithmetic — PlanetMath](https://planetmath.org/cardinalarithmetic)
8. [Introduction to Cardinal Arithmetic (Holz, Steffens, Weitz, Springer)](https://link.springer.com/book/10.1007/978-3-0346-0330-0)
9. [Refinement postulate and cancellation law for surjective cardinals (arXiv, August 2024)](https://arxiv.org/pdf/2408.04287)

---
*Topic: Encyclopedia › Physical world and mathematics › Mathematics and statistics › Numbers and algebra › Arithmetic and number systems › Number systems › Ordinal and cardinal numbers › Cardinal numbers › Cardinal arithmetic*

*Initially written Sep 17, 2026 · Reviewed: Sep 30, 2026 · Edited: Sep 30, 2026; Oct 11, 2026 · Last review: Sep 30, 2026*

*Copyright 2026 EdgeChat AI, a subsidiary of Biostate AI.*

License: Edgepedia Community License 1.0, https://www.edgepedia.org/edgepedia/license