Fact-checked by Grok 2 weeks ago

Measure (mathematics)

In mathematics, measure theory is a branch of real analysis that provides a rigorous framework for assigning a notion of "size," length, area, or volume to subsets of a given space, generalizing intuitive geometric measures to abstract sets.^[1] It formalizes the concept of a measure as a countably additive set function μ defined on a σ-algebra of subsets of a set X, where μ maps to the extended non-negative reals [0, ∞], with μ(∅) = 0 and additivity over disjoint countable unions.^[2] The theory originated in the early 20th century through the work of French mathematician Henri Lebesgue, who introduced the foundational ideas in his 1902 doctoral dissertation Intégrale, longueur, aire, building on prior contributions from Émile Borel and others to resolve limitations in Riemann integration.^[3]^[4] Central to measure theory are σ-algebras, collections of subsets closed under complementation and countable unions, which define the domain of measurable sets on which a measure operates consistently.^[5] Measurability extends to functions, where a function f: X → ℝ is measurable if the preimage of every Borel set is measurable, enabling the definition of integrals as limits of simple function approximations.^[6] The Lebesgue measure, a specific complete measure on ℝⁿ that coincides with Euclidean length, area, and volume for intervals and rectangles, is constructed via outer measure (the infimum of coverings by intervals) and Carathéodory's criterion for measurability, ensuring countable additivity on the resulting σ-algebra.^[7] This construction, due to Lebesgue in 1902, allows handling of "pathological" sets like the Vitali set, which are non-measurable under the axiom of choice.^[8] Measure theory underpins modern integration, with the Lebesgue integral extending the Riemann integral to a broader class of functions (e.g., bounded functions on sets of finite measure, or unbounded functions via limits), and dominating convergence theorems that facilitate interchanging limits and integrals.^[4] In probability theory, measures normalize to probability measures (with total mass 1), providing the axiomatic foundation for Kolmogorov's probability spaces and enabling rigorous treatment of random variables and expectations.^[1] Applications extend to functional analysis (e.g., Lᵖ spaces), partial differential equations, and ergodic theory, while generalizations like Hausdorff measure address fractal dimensions and geometric properties.^[9] The theory's emphasis on null sets (measure zero) allows quotienting by negligible differences, unifying continuous and discrete mathematics.^[2]

Foundations

Definition

In measure theory, the foundational structure for defining a measure is a sigma-algebra on a set X, which is a collection \mathcal{A} of subsets of X (called measurable sets) that includes the empty set \emptyset and X itself, and is closed under complements and countable unions (and hence also countable intersections).^[10] This ensures that the family of measurable sets is sufficiently rich to support operations needed for measuring sizes in a consistent manner.^[11] A measure \mu on a measurable space (X, \mathcal{A}) is formally defined as a function \mu: \mathcal{A} \to [0, \infty] satisfying two key axioms: \mu(\emptyset) = 0, and for any countable collection of pairwise disjoint sets \{A_n\}_{n=1}^\infty \subset \mathcal{A},

\mu\left( \bigcup_{n=1}^\infty A_n \right) = \sum_{n=1}^\infty \mu(A_n).

^[11] The non-negativity axiom requires that \mu(A) \geq 0 for all A \in \mathcal{A}, while the null empty set axiom specifies \mu(\emptyset) = 0.^[10] The countable additivity axiom, also known as \sigma-additivity, extends finite additivity to countable disjoint unions, allowing measures to handle infinite processes appropriately.^[11] Measures are typically extended real-valued, meaning they can take the value \infty for certain sets, which accommodates unbounded spaces; in contrast, positive measures are sometimes restricted to finite values, though the extended version is standard in general measure theory.^[10] A classic example is the Lebesgue measure on \mathbb{R}, which assigns lengths to intervals.^[11]

Simple Instances

The counting measure on a set X is defined on the power set \sigma-algebra $2^X by \mu(A) = |A| if A \subseteq X is finite and \mu(A) = \infty otherwise.^[11] This construction satisfies the axioms of a measure, as the empty set has measure zero and countable disjoint unions add up correctly, with infinite sets receiving infinite measure.^[11] For instance, on the natural numbers \mathbb{N}, it assigns measure 1 to singletons and \infty to infinite subsets. The Dirac measure \delta_x centered at a point x \in X is defined on any \sigma-algebra containing the singletons by \delta_x(A) = 1 if x \in A and $0 otherwise.^[10] It is a simple example of a measure concentrated at a single point, verifying the measure axioms through the indicator function $1_{\{x\}}, and extends to probability measures when normalized, though here it totals 1 without further scaling.^[10] Lebesgue measure on \mathbb{R}^n arises from the outer measure construction, where the outer measure \lambda^*(E) of a set E is the infimum of sums of volumes of countable rectangular coverings of E.^[12] Measurable sets are then those satisfying Carathéodory's criterion: for any set T, \lambda^*(T) = \lambda^*(T \cap E) + \lambda^*(T \setminus E), restricting the outer measure to form a complete measure on the Lebesgue \sigma-algebra.^[12] In one dimension, it assigns \lambda([a,b]) = b - a for closed intervals, extending to Borel sets and beyond. Haar measure on a locally compact topological group G is a left-invariant measure (unique up to positive scalar multiple) existing by the Riesz representation theorem, which associates it to positive linear functionals on continuous functions with compact support.^[13] For the additive group \mathbb{R}, it coincides with Lebesgue measure, while on the discrete group \mathbb{Z}, it recovers the counting measure.^[14] This invariance ensures \mu(gA) = \mu(A) for g \in G and measurable A.^[13]

Core Properties

Monotonicity and Additivity

Countable additivity implies finite additivity for a measure \mu on a \sigma-algebra. Specifically, if \{A_i\}_{i=1}^n is a finite collection of pairwise disjoint measurable sets, then \mu\left(\bigcup_{i=1}^n A_i\right) = \sum_{i=1}^n \mu(A_i). This follows by extending the finite collection to a countable one with empty sets: set A_k = \emptyset for k > n, so \mu\left(\bigcup_{i=1}^\infty A_i\right) = \sum_{i=1}^\infty \mu(A_i) = \sum_{i=1}^n \mu(A_i) + \sum_{k=n+1}^\infty \mu(\emptyset) = \sum_{i=1}^n \mu(A_i), since \mu(\emptyset) = 0.^[7] A direct consequence of finite additivity and non-negativity of measures is monotonicity: if A \subseteq B are measurable sets, then \mu(A) \leq \mu(B). To see this, note that B = A \cup (B \setminus A) where A and B \setminus A are disjoint, so \mu(B) = \mu(A) + \mu(B \setminus A) \geq \mu(A) because \mu(B \setminus A) \geq 0.^[7] For non-disjoint sets, finite additivity extends to general finite unions via disjoint decomposition. For two measurable sets A and B, decompose A \cup B = A \cup (B \setminus A), yielding \mu(A \cup B) = \mu(A) + \mu(B \setminus A). Since B \setminus A \subseteq B, monotonicity implies \mu(B \setminus A) \leq \mu(B), so \mu(A \cup B) \leq \mu(A) + \mu(B); this is finite subadditivity. Equivalently, using the identity A \cup B = (A \setminus B) \cup (B \setminus A) \cup (A \cap B) with disjoint parts, \mu(A \cup B) = \mu(A \setminus B) + \mu(B \setminus A) + \mu(A \cap B), which rearranges via monotonicity and additivity to the inclusion-exclusion formula \mu(A \cup B) = \mu(A) + \mu(B) - \mu(A \cap B). These relations hold for any finite number of sets by iterative decomposition.^[7] On the power set of an infinite set, finitely additive measures exist that are not countably additive, often constructed using the axiom of choice via ultrafilters; a simple example arises on the subalgebra of finite and cofinite sets (the finite-cofinite algebra), where one defines \mu(E) = 0 if E is finite and \mu(E) = 1 if E is cofinite. This \mu is finitely additive: for disjoint E_1, \dots, E_n in the algebra, their union is finite if all are finite (so \mu(\bigcup E_i) = 0 = \sum \mu(E_i)) or cofinite if at least one is cofinite (since the complement of the union is the intersection of complements, finite if any is cofinite, hence \mu(\bigcup E_i) = 1 = \sum \mu(E_i) as exactly one term is 1 and others 0). However, it fails countable additivity on countable disjoint finite sets covering the space.^[15]

Continuity and Subadditivity

In measure theory, a key extension of finite additivity is countable subadditivity, which states that for any countable collection of measurable sets \{A_n\}_{n=1}^\infty in a measure space (X, \mathcal{M}, \mu), the measure of their union satisfies \mu\left(\bigcup_{n=1}^\infty A_n\right) \leq \sum_{n=1}^\infty \mu(A_n).^[16] This inequality follows from the monotonicity and countable additivity of \mu: first, construct a disjoint collection \{B_n\}_{n=1}^\infty such that \bigcup_{n=1}^\infty B_n = \bigcup_{n=1}^\infty A_n by setting B_1 = A_1 and B_n = A_n \setminus \bigcup_{k=1}^{n-1} A_k for n \geq 2; monotonicity implies \mu(B_n) \leq \mu(A_n) for each n, and countable additivity yields \mu\left(\bigcup_{n=1}^\infty A_n\right) = \mu\left(\bigcup_{n=1}^\infty B_n\right) = \sum_{n=1}^\infty \mu(B_n) \leq \sum_{n=1}^\infty \mu(A_n).^[16] Countable subadditivity enables the analysis of limits of sets, leading to continuity properties of measures. Specifically, continuity from below holds: if \{A_n\}_{n=1}^\infty is an increasing sequence of measurable sets (i.e., A_n \uparrow A where A = \bigcup_{n=1}^\infty A_n), then \lim_{n \to \infty} \mu(A_n) = \mu(A). To see this, define D_1 = A_1 and D_n = A_n \setminus A_{n-1} for n \geq 2; then the D_n are disjoint, \bigcup_{n=1}^\infty D_n = A, and A_n = \bigcup_{k=1}^n D_k, so \mu(A_n) = \sum_{k=1}^n \mu(D_k) and \mu(A) = \sum_{k=1}^\infty \mu(D_k) = \lim_{n \to \infty} \mu(A_n) by countable additivity.^[16]^[17] Dually, continuity from above applies to decreasing sequences: if \{A_n\}_{n=1}^\infty is decreasing with A_n \downarrow A where A = \bigcap_{n=1}^\infty A_n and \mu(A_1) < \infty, then \lim_{n \to \infty} \mu(A_n) = \mu(A).^[16] The proof relies on complements: consider the increasing sequence A_1 \setminus A_n \uparrow A_1 \setminus A, which has measure \mu(A_1) - \mu(A_n); by continuity from below, \lim_{n \to \infty} \mu(A_1 \setminus A_n) = \mu(A_1 \setminus A) = \mu(A_1) - \mu(A), so subtracting from \mu(A_1) yields the result, with the finite measure condition propagating through the differences.^[17] These continuity properties thus connect the measures of limiting sets directly to the limits of their measures, underpinning approximations in integration and probability.

Advanced Properties

Completeness and Regularity

A measure space (X, \mathcal{M}, \mu) is called complete if every subset of a null set (a measurable set of measure zero) is itself measurable and hence also null.^[18] This property ensures that the sigma-algebra \mathcal{M} includes all subsets of sets with measure zero, preventing "invisible" non-measurable subsets within null sets. Completeness simplifies many arguments in analysis by allowing subsets of null sets to be treated as measurable without altering the measure. The completion of a measure space addresses incompleteness by extending the sigma-algebra to include all subsets of null sets. Specifically, for a measure space (X, \mathcal{M}, \mu), the completion \overline{\mathcal{M}} consists of all sets of the form A \cup N where A \in \mathcal{M} and N \subseteq B for some B \in \mathcal{M} with \mu(B) = 0, or equivalently A \Delta N with A \in \mathcal{M} and N null. The extended measure \overline{\mu} is defined by \overline{\mu}(A \cup N) = \mu(A), which preserves the original measure on \mathcal{M} and assigns measure zero to all new sets. This construction yields a complete measure space (\overline{X}, \overline{\mathcal{M}}, \overline{\mu}) that is minimal in the sense that it is the smallest complete extension containing the original sigma-algebra.^[18] The Lebesgue measure on \mathbb{R}^n becomes complete upon this completion process.^[19] Regularity properties provide ways to approximate measurable sets using simpler topological sets, refining the structure of measures on topological spaces. A measure \mu on a topological space is outer regular if for every measurable set A,

\mu(A) = \inf \{ \mu(U) : U \supseteq A, \, U \text{ open} \},

allowing approximation from above by open sets. Similarly, \mu is inner regular if

\mu(A) = \sup \{ \mu(K) : K \subseteq A, \, K \text{ compact} \},

enabling approximation from below by compact sets. These properties hold for the Lebesgue measure on Borel sets in \mathbb{R}^n, where outer regularity applies to all subsets via the outer measure, and inner regularity holds for measurable sets.^[12] In the context of Borel measures on locally compact Hausdorff spaces, Radon measures exemplify strong regularity. A Radon measure is a Borel measure that is finite on compact sets, outer regular on all Borel sets, and inner regular on open sets. For such measures, the inner regularity extends to all Borel sets under sigma-finiteness, and theorems guarantee that Borel regular outer measures (outer regular and finite on compacts) coincide with Radon measures. This "dropping the edge" phenomenon allows precise approximation of Borel sets by open or closed sets, crucial for integration and duality in functional analysis.^[20] In probability theory, inner regularity manifests as tightness: a probability measure \mu on a metric space is tight if for every Borel set A and \epsilon > 0, there exists a compact K \subseteq A with \mu(A \setminus K) < \epsilon, equivalent to \mu(A) = \sup \{ \mu(K) : K \subseteq A, \, K \text{ compact} \}. Thus, tightness is precisely the inner regularity condition for probability measures, ensuring no mass escapes to infinity and facilitating weak convergence results like Prokhorov's theorem.^[21]

Finite and Sigma-Finite Measures

A measure \mu on a measurable space (X, \Sigma) is finite if \mu(X) < \infty.^[22] Finite measures exhibit robust properties, including continuity from above: for any decreasing sequence of measurable sets A_n \downarrow A, it holds that \lim_{n \to \infty} \mu(A_n) = \mu(A).^[16] A measure \mu is \sigma-finite if X = \bigcup_{n=1}^\infty X_n for some sequence of measurable sets \{X_n\} with \mu(X_n) < \infty for each n.[](https://e.math.cornell.edu/people/belk/measure theory/MoreMeasureTheory.pdf) This condition permits the total measure \mu(X) to be infinite while allowing decomposition into countably many finite-measure components, facilitating the extension of finite-measure techniques to broader settings. For instance, Lebesgue measure \lambda on \mathbb{R} (with the Borel \sigma-algebra) is \sigma-finite, as

\mathbb{R} = \bigcup_{n \in \mathbb{Z}} [n, n+1)&#36; and

\lambda([n, n+1)) = 1 < \inftyfor eachn. In contrast, counting measure on an uncountable set like \mathbb{R}(where\mu(A) = |A|ifAis finite and\inftyotherwise) is not\sigma-finite, since sets of finite measure are precisely the finite subsets, and any countable union of finite sets remains countable, failing to cover \mathbb{R}$.[](https://e.math.cornell.edu/people/belk/measure theory/MoreMeasureTheory.pdf) \sigma-Finiteness enhances continuity properties beyond those of general measures: for A_n \downarrow A with each A_n \in \Sigma, \lim_{n \to \infty} \mu(A_n) = \mu(A) holds without requiring \mu(A_1) < \infty, as the finite decomposition allows restriction to finite-measure portions where standard continuity applies.^[23] This assumption is also essential for theorems like Fubini's, which equates the integral of a nonnegative measurable function over a product space to iterated integrals when both measures are \sigma-finite.^[24] Every \sigma-finite measure is semifinite.^[25]

Semifinite and Localizable Measures

A semifinite measure on a measurable space (X, \mathcal{A}) is defined such that for every set E \in \mathcal{A} with \mu(E) > 0, there exists a subset F \subseteq E in \mathcal{A} satisfying $0 < \mu(F) < \infty.^[26] This condition ensures that sets of infinite measure can be "approximated" by subsets of finite positive measure. An equivalent characterization is that the measure has no infinite atoms, meaning there is no set E with \mu(E) = \infty such that every measurable subset of E has measure either 0 or \infty.^[27] All \sigma-finite measures are semifinite, as they can be exhausted by countable unions of finite-measure sets, allowing extraction of finite subsets from positive-measure sets.^[28] A basic example of a semifinite but not \sigma-finite measure is the counting measure on an uncountable set X, where \mu(A) = |A| if A is finite and \infty otherwise; singletons have measure 1, so any non-empty set contains a finite positive-measure subset.^[26] Another example is Dieudonné's measure on the first uncountable ordinal \omega_1 equipped with its order topology, a semifinite Borel measure on this non-locally compact space where measures of initial segments grow without bound but finite subsets exist for positive sets.^[29] An involved example arises in uncountable products: consider the product space \prod_{i \in I} \mathbb{N} for uncountable I, endowed with the product \sigma-algebra and the product of counting measures on each factor; this yields a semifinite measure, as finite-support sections provide finite-measure subsets within positive sets, though the total space has infinite measure without \sigma-finiteness.^[30] Non-examples of semifinite measures include the pathological measure \mu(A) = 0 if A = \emptyset and \mu(A) = \infty otherwise on the power set of any non-empty set; here, every non-empty set has infinite measure, but no subset has finite positive measure, violating the condition.^[27] Semifinite measures support integral decompositions where the integral of a non-negative measurable function f is the supremum of integrals over finite-measure subsets, enabling extension of integration theory beyond \sigma-finiteness while avoiding pathologies in infinite cases.^[31] Localizable measures provide a stricter framework: a semifinite measure is localizable if the measurable space admits a directed family of finite submeasures whose pointwise supremum recovers the original measure, ensuring Dedekind completeness in the measure algebra for robust handling of suprema over disjoint families.^[32] Localizability implies semifiniteness but is not equivalent, as some semifinite measures like certain pathological products fail the directed exhaustion property; \sigma-finite measures are special cases of localizable ones.^[33]^[34]

Pathological and Limiting Cases

Non-Measurable Sets

In measure theory, particularly for the Lebesgue measure on the real numbers, non-measurable sets are subsets that cannot be assigned a measure value while preserving the axioms of additivity, monotonicity, and translation invariance. These sets highlight fundamental limitations in extending measure from simple intervals to all subsets of the space, as the collection of measurable sets forms a proper sigma-algebra that excludes certain pathological subsets. The existence of such sets underscores the incompleteness of the Lebesgue sigma-algebra and the role of set-theoretic assumptions in determining what can be measured. A canonical example is the Vitali set, constructed within the unit interval [0,1] using the axiom of choice. Consider the equivalence relation on [0,1] where two points x, y \in [0,1] are equivalent if x - y \in \mathbb{Q}. This partitions [0,1] into uncountably many equivalence classes, each dense in [0,1]. Selecting exactly one representative from each class via the axiom of choice yields a set V \subset [0,1], known as a Vitali set. The rational translates V + q = \{v + q \mid v \in V\} for q \in \mathbb{Q} \cap [0,1) are pairwise disjoint, and their union covers [0,1]. If V were Lebesgue measurable with measure \mu(V) = m > 0, then the measure of the union would be countably infinite, exceeding the measure of [0,1], which is 1; if m = 0, the union would have measure 0, again contradicting the coverage of [0,1]. Thus, V is non-measurable. This construction was first given by Giuseppe Vitali in 1905.^[35] In higher dimensions, non-measurable sets appear in the Banach-Tarski paradox, which decomposes the unit ball in \mathbb{R}^3 into finitely many non-measurable pieces that can be rigidly reassembled into two copies of the original ball. The proof relies on the axiom of choice to select representatives from cosets of a free subgroup of rank 2 in the special orthogonal group SO(3), enabling paradoxical rotations that double the volume without stretching. Specifically, the ball is partitioned into pieces equivariant under these group actions, each piece non-measurable with respect to Lebesgue measure in \mathbb{R}^3. This result, established by Stefan Banach and Alfred Tarski in 1924, extends the idea of non-measurability to geometric decompositions and illustrates how choice enables counterintuitive equidissections in Euclidean spaces of dimension at least 3.^[36] The existence of non-measurable sets is tied to the axiom of choice (AC); without it, such sets may not be provable. In Zermelo-Fraenkel set theory (ZF) augmented by the axiom of dependent choice (DC), it is consistent that all subsets of the reals are Lebesgue measurable. Robert Solovay constructed such a model in 1970, assuming the existence of a strongly inaccessible cardinal, where every set of reals has the property of Baire, is Lebesgue measurable, and has the perfect set property. However, in full Zermelo-Fraenkel set theory with choice (ZFC), non-measurable sets necessarily exist, as shown by constructions like the Vitali set.^[37] Non-measurable sets can still be assigned a Lebesgue outer measure, defined as the infimum of the total length of countable open interval covers. For the Vitali set V \subset [0,1], the outer measure is 1, matching that of [0,1], since any cover must encompass the entire interval due to the density of its translates. However, the inner measure, approximated from below by compact subsets, is 0, as no positive measure compact set can intersect all equivalence classes without overlapping rationals improperly. This discrepancy—positive outer measure but no exact measure—prevents inclusion in the Lebesgue sigma-algebra. The historical discovery of non-measurable sets traces to Vitali's 1905 work, which resolved the open problem of whether Lebesgue measure extends to all subsets by showing it does not under standard axioms.^[35]

Infinite and s-Finite Measures

Infinite measures arise in measure theory when the total measure of the underlying space is infinite, yet the measure still adheres to the fundamental axioms: non-negativity, μ(∅) = 0, and countable subadditivity for disjoint sets. These measures are essential for modeling unbounded spaces, such as the real line under Lebesgue measure, where μ(ℝ) = ∞. A simple example is the counting measure on the natural numbers ℕ equipped with the power set σ-algebra, defined by μ(A) = |A| if A is finite and μ(A) = ∞ otherwise; this satisfies countable additivity since infinite disjoint unions yield infinity.^[22] s-Finite measures provide a framework for handling certain infinite measures by decomposing them into countable sums of finite measures. Specifically, a measure μ on (X, Σ) is s-finite if there exist finite measures μ_n (n ∈ ℕ) such that μ = ∑{n=1}^∞ μ_n, where the sum is defined pointwise on sets: μ(E) = ∑{n=1}^∞ μ_n(E) for E ∈ Σ. Every σ-finite measure is s-finite, as it can be expressed via restrictions to the finite-measure sets in its decomposition, but the converse fails; for instance, the measure on a singleton space {x} with μ({x}) = ∞ and μ(∅) = 0 is s-finite (e.g., as ∑ n δ_x where δ_x is the Dirac measure of mass 1), yet not σ-finite since no proper nonempty subset has finite positive measure.^[38]^[38] Properties of s-finite measures include partial extensions of integration results, such as limited Fubini-type theorems for products where one factor is finite, though full product measure constructions and unrestricted iterated integrals generally require σ-finiteness to ensure the product σ-algebra is properly defined and additivity holds without anomalies. Unlike semifinite measures, which ensure every positive-measure set contains a finite-measure subset, s-finite measures need not satisfy this; the singleton example above illustrates that s-finiteness permits "purely infinite" components without finite approximations. In infinite groups, left-invariant Haar measures can be infinite and s-finite, as seen in certain noncompact Lie groups where the measure decomposes into countable finite parts, facilitating ergodic theory applications despite the overall infinity.^[38] A key limitation of infinite and s-finite measures is the potential absence of continuity from above: if {E_n} is a decreasing sequence of measurable sets with ∩ E_n = ∅, then μ(E_n) may not converge to 0 without additional conditions like σ-finiteness or finite measure on the E_n. For s-finite measures, this continuity can hold on the supports of the finite components but fails globally in pathological cases, underscoring the need for stricter assumptions in theorems involving limits.

Generalizations

Signed and Complex Measures

A signed measure on a measurable space (X, \mathcal{M}) is a function \nu: \mathcal{M} \to [-\infty, \infty] that is countably additive and satisfies \nu(\emptyset) = 0, with the additional property that \nu takes at most one infinite value (either +\infty or -\infty, but not both).^[39] Unlike positive measures, signed measures can take negative values, but for the finite case, the total variation |\nu|(X) < \infty.^[40] This extension builds on positive measures by allowing the codomain to be the extended reals while preserving countable additivity.^[41] The Jordan decomposition theorem provides a canonical way to express any signed measure \nu as the difference of two positive measures: \nu = \nu^+ - \nu^-, where \nu^+ and \nu^- are mutually singular, meaning there exists a set E \in \mathcal{M} such that \nu^+(X \setminus E) = 0 and \nu^-(E) = 0.^[42] The positive part is defined as \nu^+(A) = \sup\{\nu(F) : F \subseteq A, F \in \mathcal{M}\} and the negative part as \nu^-(A) = -\inf\{\nu(F) : F \subseteq A, F \in \mathcal{M}\}, ensuring uniqueness of the decomposition.^[40] The total variation of \nu is the positive measure |\nu|(A) = \nu^+(A) + \nu^-(A), which satisfies |\nu|(A) = \sup\left\{\sum_{i=1}^n |\nu(A_i)| : \{A_i\}_{i=1}^n \text{ is a partition of } A\right\}.^[39] A basic example of a signed measure is the difference of two positive measures, such as \nu(A) = \mu_1(A) - \mu_2(A) where \mu_1 and \mu_2 are positive and mutually singular on disjoint supports.^[43] More generally, the Riesz representation theorem establishes that every continuous linear functional on the space of continuous functions with compact support C_c(X) on a locally compact Hausdorff space X corresponds to integration against a regular signed (or complex) Borel measure.^[44] Complex measures generalize signed measures further by taking values in \mathbb{C}, defined as \mu: \mathcal{M} \to \mathbb{C} with countable additivity and \mu(\emptyset) = 0, where the total variation |\mu| is a positive finite measure given by |\mu|(A) = \sup\left\{\sum_{i=1}^n |\mu(A_i)| : \{A_i\}_{i=1}^n \text{ partitions } A\right\}.^[45] Any complex measure decomposes as \mu = \mu_r + i \mu_i with real and imaginary parts as signed measures, and its total variation satisfies |\mu|(X) < \infty.^[42] For signed measures, absolute continuity and singularity extend naturally: a signed measure \nu is absolutely continuous with respect to a positive measure \mu if |\nu|(A) = 0 whenever \mu(A) = 0, and two signed measures \nu_1, \nu_2 are singular if there exists E \in \mathcal{M} such that |\nu_1|(X \setminus E) = 0 and |\nu_2|(E) = 0.^[43] These properties underpin the Radon-Nikodym theorem for signed measures, where absolute continuity implies \nu is representable as integration against an L^1(\mu)-function.^[46]

Vector and Finitely Additive Measures

Finitely additive measures extend the classical concept of measures by requiring additivity only over finite disjoint unions of sets, rather than countable ones. Formally, given an algebra \mathcal{A} of subsets of a set X, a finitely additive measure \mu: \mathcal{A} \to [0, \infty] satisfies \mu(\emptyset) = 0 and \mu(A \cup B) = \mu(A) + \mu(B) whenever A, B \in \mathcal{A} are disjoint, with \mathcal{A} closed under finite unions and complements but not necessarily countable operations.^[10] Unlike \sigma-additive measures, finitely additive ones are defined on algebras rather than \sigma-algebras, allowing broader applicability but potentially leading to pathologies such as non-measurable sets under the axiom of choice.^[47] A prominent example of a finitely additive measure is the Banach limit on the space \ell^\infty of bounded real sequences, which extends the standard limit functional via the Hahn-Banach theorem and induces a translation-invariant probability measure on the power set of \mathbb{N} by setting \mu(A) = L(\chi_A), where \chi_A is the characteristic sequence of A and L is the Banach limit satisfying \liminf x_n \leq L(x) \leq \limsup x_n for any x \in \ell^\infty.^[48] Another construction uses non-principal ultrafilters on \mathbb{N}: for a free ultrafilter \mathcal{U}, define \mu(A) = 1 if A \in \mathcal{U} and $0otherwise, yielding a finitely additive{0,1}-valued [probability measure](/page/Probability_measure) on \mathcal{P}(\mathbb{N})that extends the asymptotic density where possible but vanishes on finite sets.[52] Such measures can be extended from subalgebras, like the finite-cofinite algebra, to the full power set using the Hahn-Banach theorem, providing finitely additive extensions that are not\sigma$-additive. Vector measures generalize scalar measures by taking values in a Banach space E, maintaining countable additivity on a \sigma-algebra \Sigma over X: a map \nu: \Sigma \to E is a vector measure if \nu(\emptyset) = 0 and \nu(\bigcup_{n=1}^\infty A_n) = \sum_{n=1}^\infty \nu(A_n) for disjoint A_n \in \Sigma, with the series converging in the norm topology of E.^[47] Integration theory for vector measures relies on concepts like Bochner and Pettis integrability: a function f: X \to E is Bochner integrable with respect to a scalar measure \mu if it is strongly measurable (almost separably valued with preimages of open sets measurable) and \int \|f\| \, d\mu < \infty, defining the indefinite integral \nu(A) = \int_A f \, d\mu as a vector measure of bounded variation; Pettis integrability weakens this to weak measurability (scalar integrals \int_A \langle f, x^* \rangle \, d\mu exist for all x^* \in E^*) and \sigma-additivity of the range in the weak topology, allowing integration in spaces where strong measurability fails, such as L^\infty functions.^[49]^[47] Key properties of vector measures include the lack of a full Fubini-Tonelli theorem without scalar restrictions: product measures may not decompose integrals over products straightforwardly, as the range may not permit unconditional convergence or slicing without additional separability assumptions on E.^[47] For instance, in geometry, the vector area measure on subsets of \mathbb{R}^2 assigns to a region E the vector \nu(E) = \left( \int_E y \, dx, -\int_E x \, dy \right), representing signed area contributions in \mathbb{R}^2, which arises as the indefinite integral of a vector-valued density and captures oriented content for applications like moment calculations.^[50] Finitely additive vector measures further relax countable additivity, often constructed via Hahn-Banach extensions or ultrafilter limits on algebras, relating back to scalar cases like signed measures where E = \mathbb{R}, but enabling applications in non-separable spaces.^[47]^[51]

References

[1]
Measure Theory Basics - UC Berkeley Statistics
Aug 24, 2023 · Measure theory is an area of mathematics concerned with measuring the “size” of subsets of a certain set.
[2]
[PDF] MEASURES
1. Fundamentals. A measure is a mathematical object that quantifies the size of sets. Each distinct measure embodies a different way to assess how big a set is ...
[3]
Henri Lebesgue (1875 - 1941) - Biography - MacTutor
Building on the work of others, including that of Émile Borel and Camille Jordan, Lebesgue formulated the theory of measure in 1901 and in his famous paper Sur ...Missing: original | Show results with:original
[4]
[PDF] applications of lebesgue measure to the cantor set and non ...
Henri Lebesgue1 (1875-1941), in the process of developing his revolution- ary definition of integration, created the concept of Lebesgue measure. between 1899 ...
[5]
https://people.math.binghamton.edu/mfochler/bu-mfx/bu-math-talks/math-club-intro-measure-theory.html
[6]
[PDF] 1 Measure Theory - Princeton University
The concept of measurable functions is a natural outgrowth of the idea of measurable sets. It stands in the same relation as the concept of continuous functions ...
[7]
[PDF] Measure Theory John K. Hunter - UC Davis Math
In these notes, we develop the theory of measures first, and then define integrals.
[8]
[PDF] Math212a1411 Lebesgue measure.
Oct 14, 2014 · In today's lecture we will discuss the concept of measurability of a subset of R. We will begin with Lebesgue's (1902) definition of.
[9]
[PDF] Introduction to Geometric Measure Theory - Stanford University
... Concepts ... In this chapter we briefly review the basic theory of outer measure, which is based on. Caratheodory's definition of measurability.
[10]
[PDF] An Introduction to Measure Theory - Terry Tao
Definition 1.4.12 (Sigma algebras). Let X be a set. A σ-algebra on X is a ... measures can only measure a σ-algebra of measurable sets. In Definition ...
[11]
[PDF] Measures - UC Davis Math
A measure is a countably additive, non-negative, extended real-valued function defined on a σ-algebra. Definition 1.9. A measure µ on a measurable space (X, A) ...
[12]
[PDF] Chapter 2: Lebesgue Measure - UC Davis Math
We will obtain Lebesgue measure as the restriction of Lebesgue outer measure to Lebesgue measurable sets. The construction, due to Carathéodory, works for any.Missing: original | Show results with:original
[13]
[PDF] the haar measure - math 519
If G is a locally compact group, then there exists a left (and right) Haar measure on G which is unique up to scalar multiple. If x ∈ G, then the map φx : g 7→ ...
[14]
https://www.math.berkeley.edu/~cjdowd/haar1.pdf
[15]
[PDF] Philosophy of Probability: Problem Sets
Define P(E) to be zero if E is finite and P(E)=1 if E is infinite. Show that P is a finitely additive probability measure on A. Show that it is not countably ...
[16]
[PDF] 2.3 Basic Properties of Measures - Christopher Heil
We give the following special name to measures that have the property that every subset of a measurable set with zero measure are measurable. Definition 2.19 ( ...<|control11|><|separator|>
[17]
[PDF] LECTURE NOTES MEASURE THEORY and PROBABILITY
Jun 20, 2003 · ) ↑ µ(A), (continuity for below). (iv) If Aj ↓ A and µ(A1) < ∞, then µ(Aj) ↓ µ(A), (continuity from above). Remark 2.2. The finiteness ...
[18]
[PDF] 2.4 The Completion of a Measure - Christopher Heil
A complete measure is one such that every subset A of every null set E is measurable (Definition 2.19). Complete measures are often more convenient to work ...
[19]
[PDF] Brief Notes on Measure Theory - UC Davis Mathematics
A measure space (X,A,µ) is complete if every subset of a set of measure zero is measurable (when its measure is necessarily zero). Every measure space (X,A,µ) ...
[20]
[PDF] Borel Regular & Radon Measures
We now state and prove an important regularity property of Borel regular ... regular outer measures are in fact Radon measures. 2.9 Lemma. Let X be a ...
[21]
[PDF] The stochastic order of probability measures on ordered metric spaces
Sep 13, 2017 · The measure µ is said to be inner regular or tight if for any Borel set A and ε > 0 there exists a compact set K ⊆ A such that µ(A) − ε<µ(K). A ...
[22]
[PDF] Lecture 2 Measures
Sep 19, 2013 · Definition 2.1 (Measure). Let (S, S) be a measurable space. A ... counting measure. Clearly, µ is a finite measure if and only is S is ...
[23]
[PDF] More Measure Theory
Similarly, the natural numbers N are σ-finite with respect to counting measure. Not every measure space is σ-finite. For example, if we put counting measure.
[24]
[PDF] Math 639: Lecture 1 - Measure theory background
(Continuity from below) If A1 ⊂ A2 ⊂ ... and A = Si Ai then. Prob(Ai ) ↑ Prob(A). (Continuity from above) If A1 ⊃ A2 ⊃ ... and A = Ti Ai ...
[25]
[PDF] Chapter 5. Product Measures - UC Davis Math
For example, if R is equipped with its Borel σ-algebra, then Q × Q is a measurable rectangle in R × R. (Note that the 'sides' A, B of a measurable rectangle A × ...
[26]
[PDF] Homework 2 Solutions
(a) Suppose µ is σ-finite. Show that µ is semifinite. (b) Suppose µ is semifinite. Show that for E ∈ M with µ(E) = ∞ and any C > 0, there exists F ⊂ E.
[27]
[PDF] 2.2 Measures - Christopher Heil
Focusing on counting measure on Rd, we some ways in which counting measure is similar to Lebesgue measure or a delta measure, but more ways in which it is ...
[28]
[PDF] Measure Theory - University of Waterloo
). Therefore ν is countably additive, and thus is a complex measure. Moreover if. µ(E) = 0, then χE = 0a.e.(µ) and hence ν(E) = *(0) = 0. Thus ν µ. By the ...
[29]
[PDF] Lecture 24: Properties of Measures - ECE, IISc
Remark 1. In practice, most measures are σ-finite. Non σ-finite measures have pathological properties. Definition 1.6 (Uniform measure).
[30]
[PDF] Chapter 43 Topologies and measures II - University of Essex
Aug 21, 2015 · (c) Dieudonné's measure (411Q) is a Borel measure on ω1 which is not tight, so ω1 is certainly not a. Radon space; as it is an open set in ω1 ...
[31]
A REMARK ON TONELLΓS THEOREM ON INTEGRATION IN ...
The product measure βι x β2 may not be semifinite even when βt is sigma-finite and β2 semifinite, as Example 1 shows.<|control11|><|separator|>
[32]
On Homogeneous Measure Algebras - PNAS
MAHARAM. An example of a homogeneous measure algebra is the Boolean algebra. P(7y) of all measurable sets (mod. null sets) of an infinite product space. Qz ...Missing: semifinite | Show results with:semifinite
[33]
[PDF] Chapter 33 Maharam's theorem - University of Essex
Nov 17, 2010 · (k) Show that a homogeneous semi-finite measure algebra is σ-finite. (l) Let (X, Σ,µ) be a measure space, and A a subset of X which has a ...
[34]
[PDF] MEASURE AND INTEGRATION - ETH Zürich
This book is based on notes for the lecture course “Measure and Integration” held at ETH Zürich in the spring semester 2014. Prerequisites are the first.Missing: seminal | Show results with:seminal
[35]
Fremlin --- Measure Theory
A complete localizable non-locally-determined space; a complete locally determined non-localizable space; a complete locally determined localizable space which ...
[36]
categories of measure theory in nLab
A localizable enhanced measurable space is an enhanced measurable space that admits a faithful semifinite measure and satisfies any of the above equivalent ...
[37]
[PDF] arXiv:2108.06406v2 [math.OA] 23 Dec 2021
Dec 23, 2021 · Then (X,Σ,µ) is a complete semifinite measure space, but it is not Dedekind. ... Maharam, On homogeneous measure algebras, Proc. Nat. Acad ...
[38]
[PDF] From the beginning of set theory to Lebesgue's measure problem
Jun 5, 2019 · Indeed, in 1905 [55], using the newly discovered principle, Vitali was able to show that the measure problem had no solution by constructing a ...
[39]
[PDF] The Banach-Tarski Paradox - Harvey Mudd College Mathematics
In 1924, S. Banach and A. Tarski proved a truly remarkable theorem: given a solid ball in R3, it is possible to partition it into finitely many pieces and.
[40]
[PDF] A Model of Set-Theory in Which Every Set of Reals is Lebesgue ...
Sep 10, 2003 · We show that the existence of a non-Lebesgue measurable set cannot be proved in Zermelo-Frankel set theory (ZF) if use of the axiom of choice is.
[41]
[1810.01837] On S-Finite Measures and Kernels - arXiv
Oct 3, 2018 · In this note, we develop some of the basic theory of s-finite (measures and) kernels, a little-studied class of kernels which Staton has recently convincingly ...
[42]
[PDF] Signed Measures
We begin with the following natural definitions: Definition 4.2.2. Let µ be a signed measure on (X, A). Let P,N,M ∈ A. Then ...
[43]
[PDF] 6 — Signed Measures - UBC Math
Nov 22, 2019 · Definition 6.1 (Signed Measures). Let X be a nonempty set and M⊂P(X) be a σ-algebra. (a) A signed measure on (X, M) is a function ν : M ...
[44]
245B, notes 1: Signed measures and the Radon-Nikodym ...
Jan 4, 2009 · In this section we investigate the structure of this space, together with the closely related spaces of signed measures and finite measures.
[45]
[PDF] 7. Signed measures and complex measures - KSU Math
In this section we discuss a generalization of the notion of a measure, to the case where the values are allowed to be outside [0, ∞].
[46]
[PDF] Signed Measures and the Radon-Nikodym Theorem
Proof If ν is a signed measure then ν is absolutely continuous with respect to µ if and only if its total variation |ν| is absolutely continuous with re- spect ...
[47]
[PDF] Another Riesz Representation Theorem - UBC Math
Oct 9, 2018 · In these notes we prove (one version of) a theorem known as the Riesz Representation. Theorem. Some people also call it the Riesz–Markov ...
[48]
[PDF] Real Analysis lecture notes for MA 645/646
defines an outer measure which yields, employing Carathéodory's construction 2.2.3, a com- plete positive measure µF on a σ-algebra MF in (a, b). µF is an ...
[49]
[PDF] Differentiation Lecture 7, Following Folland, ch 3.1, 3.2
We fix a measure µ on (X,M) and decompose another signed measure ν into two parts, one is singular to µ and another is absolutely continuous with respect to µ.
[50]
AMS eBooks: Mathematical Surveys and Monographs
Vector Measures · J. Diestel · View full volume as PDF · Download chapters as PDF · Front/Back Matter · Chapters.
[51]
Vector Measures - Joseph Diestel, John Jerry Uhl - Google Books
Limited preview - 1977. Vector Measures · Joseph Diestel,John Jerry Uhl No preview available - 1977. Bibliographic information. Title, Vector Measures Volume 15 ...
[52]
[PDF] Ultrafilters, with applications to analysis, social choice and ...
Sep 3, 2009 · We define the notion of an ultrafilter on a set, and present three applications. The first is an alternative presentation of the Banach limit of ...
[53]
[PDF] PETTIS INTEGRATION
Basic books: J. Diestel, J.J. Uhl Vector measures, Math. Surveys 15(1977). K. Musiaª, Topics in the theory of Pettis integration, Rend ...
[54]
Vector Measures - Joseph Diestel, John Jerry Uhl - Google Books
In this survey the authors endeavor to give a comprehensive examination of the theory of measures having values in Banach spaces.