Fact-checked by Grok 2 weeks ago

Counting measure

In measure theory, the counting measure on a set X is defined as the function \mu that assigns to each subset E \subseteq X its cardinality |E| if E is finite, and \infty if E is infinite, with the domain being the power set \mathcal{P}(X) or any \sigma-algebra on X.^[1] This measure is particularly natural on countable sets, such as the natural numbers \mathbb{N}, where it quantifies the "size" of subsets by simply counting elements, bridging discrete mathematics and continuous measure-theoretic frameworks.^[2] Key properties of the counting measure include non-negativity (\mu(E) \geq 0 for all E), \mu(\emptyset) = 0, and countable additivity: for a countable collection of disjoint subsets \{E_i\}, \mu\left(\bigcup_i E_i\right) = \sum_i \mu(E_i), which holds even when the sum diverges to infinity.^[2] It is semifinite (every set of positive measure contains a subset of finite positive measure), and \sigma-finite if and only if X is at most countable (i.e., finite or countably infinite); it is not \sigma-finite on uncountable sets, as any countable union of finite-measure sets is at most countable.^[1] On finite sets, it coincides with the cardinality function, satisfying rules like additivity for disjoint unions (|\cup A_i| = \sum |A_i|) and monotonicity (if A \subseteq B, then |A| \leq |B|).^[3] The counting measure plays a central role in discrete probability and integration, where the integral of a function f: X \to [0, \infty) with respect to \mu reduces to the sum \int_X f \, d\mu = \sum_{x \in X} f(x) over the points in X, facilitating the study of series and mass functions on countable spaces.^[2] It contrasts with measures like Lebesgue measure on \mathbb{R}^d, which is \sigma-finite and assigns zero measure to countable sets, highlighting the counting measure's emphasis on discrete structure rather than continuum.^[1] In applications, it underpins combinatorial counting principles, such as the inclusion-exclusion formula for the cardinality of unions: |\cup A_i| = \sum |A_i| - \sum |A_i \cap A_j| + \cdots + (-1)^{n+1} |\cap A_i|.^[3]

Definition and Properties

Formal Definition

The counting measure on a set X is a measure \mu defined on the power set \mathcal{P}(X), which serves as the \sigma-algebra of all subsets of X, such that for any subset A \subseteq X, \mu(A) = |A| if A is finite and \mu(A) = \infty if A is infinite, where |A| denotes the cardinality of A.^[4]^[2] This definition assigns to each measurable set the "size" of that set in terms of its number of elements, extending naturally to infinity for infinite sets, and satisfies \mu(\emptyset) = 0.^[4] The power set \mathcal{P}(X) is always a \sigma-algebra, but in practice, the counting measure is often considered on the \sigma-algebra generated by the singletons \{x\} for x \in X, which coincides with \mathcal{P}(X) when X is at most countable.^[2] For the measure to be \sigma-finite in non-trivial cases—meaning X can be expressed as a countable union of sets of finite measure—X must be at most countable, as uncountable sets would have subsets of infinite measure that prevent such a decomposition.^[4]/01:_Foundations/1.07:_Counting_Measure) Common notations for the counting measure include \# or, when X is countable, the sum of Dirac measures \sum_{x \in X} \delta_x, where \delta_x is the Dirac measure at x, reflecting that \mu(A) counts the points in A via these point masses.^[2]/01:_Foundations/1.07:_Counting_Measure)^[5] To verify that this defines a measure, it must satisfy non-negativity, \mu(\emptyset) = 0, and countable additivity: for a countable collection of pairwise disjoint sets \{A_i\}_{i=1}^\infty \subseteq \mathcal{P}(X), \mu\left(\bigcup_{i=1}^\infty A_i\right) = \sum_{i=1}^\infty \mu(A_i).^[4]^[2] Non-negativity holds since cardinalities are non-negative, and \mu(\emptyset) = 0 by definition. For additivity, if all A_i are finite and the union is finite, the cardinalities add directly as | \bigcup A_i | = \sum |A_i|; if the union is infinite or any A_i is infinite, both sides equal \infty, preserving the equality under the extended reals [0, \infty].^[2]

Basic Properties

The counting measure \mu on a set X satisfies the monotonicity property: if A \subseteq B \subseteq X, then \mu(A) \leq \mu(B). This holds because the cardinality of A cannot exceed that of B, with finite cardinals comparing numerically and infinite cardinals yielding \infty \leq \infty.^[6] The counting measure is \sigma-finite when X is countable, as X decomposes into a countable union of singletons, each with finite measure \mu(\{x\}) = 1. However, on an uncountable set X, the counting measure is not \sigma-finite, since sets of finite measure are precisely the finite subsets, and any countable union of such sets remains countable, failing to cover X while keeping all components of finite measure; consequently, \mu(X) = \infty.^[6]^[1] The measure space (X, \mathcal{P}(X), \mu) with the power set \sigma-algebra is complete. Null sets are those with measure zero, but since \mu(A) = 0 if and only if A = \emptyset (as every non-empty set contains at least one point with measure 1), there are no non-trivial null sets. Thus, every subset of a null set is measurable, satisfying the completeness condition.^[7] The counting measure arises as an extension of point masses via outer measure construction. Specifically, it is the measure obtained from the sum of Dirac measures \delta_x at each point x \in X, where \delta_x(E) = 1 if x \in E and 0 otherwise, yielding \mu(E) = \sum_{x \in X} \delta_x(E) for measurable E.^[6] For a countable collection of pairwise disjoint measurable sets \{A_n\}_{n=1}^\infty \subseteq X,

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

with the sum equaling the cardinality of the disjoint union \bigsqcup_{n=1}^\infty A_n. This countable additivity follows from the \sigma-additivity axiom, as the measure aligns with cardinal summation under disjointness.^[6]

Examples and Constructions

On Natural Numbers

The counting measure on the natural numbers \mathbb{N} (assuming \mathbb{N} = \{1, 2, 3, \dots \}) is defined with respect to the power set \mathcal{P}(\mathbb{N}) as the \sigma-algebra, where the measure \mu assigns \mu(\{n\}) = 1 to each singleton \{n\} for n \in \mathbb{N}, and extends countably additively to all subsets./01%3A_Foundations/1.07%3A_Counting_Measure)^[2] For any finite subset A \subset \mathbb{N}, \mu(A) equals the cardinality |A|, reflecting the number of elements in A. For instance, \mu(\{1, 3, 5\}) = 3.^[8] In contrast, any infinite subset B \subset \mathbb{N} receives measure \mu(B) = \infty, including the full set \mu(\mathbb{N}) = \infty./01%3A_Foundations/1.07%3A_Counting_Measure) This behavior highlights the distinction between finite and infinite subsets under the counting measure. For example, the cofinite set \mathbb{N} \setminus \{1\} is infinite and thus has \mu(\mathbb{N} \setminus \{1\}) = \infty, whereas its finite complement \{1\} has measure 1.^[2] Finite complements of infinite sets yield finite measures only if the infinite set is cofinite (i.e., has a finite complement). In such cases, the cofinite infinite set has infinite measure, while its finite complement has finite measure, emphasizing the measure's sensitivity to cardinality rather than density or other properties.^[8] The power set \mathcal{P}(\mathbb{N}) as the \sigma-algebra aligns with the Borel \sigma-algebra generated by the discrete topology on \mathbb{N}, where every subset is open./01%3A_Foundations/1.11%3A_Measurable_Spaces) This topology is induced by the discrete metric d(m, n) = 1 if m \neq n and d(m, m) = 0, making \mathbb{N} a discrete metric space compatible with the counting measure's structure./01%3A_Foundations/1.11%3A_Measurable_Spaces)

On General Countable Sets

The counting measure can be extended to any countable set X, which may be enumerated as X = \{x_1, x_2, \dots \}, by defining the measure \mu on subsets A \subseteq X as \mu(A) = \sum_{x \in A} 1, which equals the cardinality of A if A is finite and \infty otherwise.^[2]^[6] This construction assigns measure 1 to each singleton \{x\} for x \in X and extends additively to finite unions, while infinite subsets receive infinite measure, reflecting the discrete nature of the space.^[4] A key property of this construction is its invariance under bijections between countable sets. Specifically, if f: X \to Y is a bijection between countable sets X and Y, then the counting measure \mu_Y on Y satisfies \mu_Y(B) = \mu_X(f^{-1}(B)) for any B \subseteq Y, as both sides equal the cardinality of B (or \infty if infinite).^[6] This invariance underscores the measure's dependence solely on combinatorial cardinality, independent of the specific labeling of elements in the set. For example, on the integers \mathbb{Z}, the counting measure assigns infinite measure to the set of even integers, since it is countably infinite.^[2] Similarly, on the rational numbers \mathbb{Q}, the subset \mathbb{Q} \cap [0,1] receives infinite measure under the counting measure, despite being dense in [0,1] and having Lebesgue measure zero; this highlights how the counting measure prioritizes pointwise enumeration over topological density.^[9] While the focus here is on countable sets, where the counting measure is \sigma-finite (as X is a countable union of singletons, each of finite measure 1), it is worth noting briefly that on uncountable sets, assigning measure 1 to singletons leads to a measure that is not \sigma-finite, with the total space having infinite measure.^[10]^[4] The counting measure on a countable set X admits a useful representation as the sum \mu = \sum_{x \in X} \delta_x, where \delta_x denotes the Dirac measure at x, which assigns 1 to sets containing x and 0 otherwise.^[6] This decomposition expresses the measure as a countable superposition of point masses, facilitating analysis in broader measure-theoretic contexts.

Integration Theory

Lebesgue Integral with Respect to Counting Measure

The Lebesgue integral with respect to the counting measure \mu on a measurable space (X, \mathcal{A}) begins with the case of indicator functions. For a measurable set A \in \mathcal{A}, the integral of the indicator function $1_A is \int 1_A \, d\mu = \mu(A), which equals the cardinality |A| if A is finite and \infty otherwise.^[11] This follows directly from the definition of the counting measure, where singletons have measure 1 and the measure is additive over disjoint sets.^[12] For simple functions, which are finite linear combinations of indicator functions, the integral extends linearly. Consider a simple function f = \sum_{k=1}^n c_k 1_{A_k}, where c_k \geq 0 are constants and the A_k \in \mathcal{A} are measurable sets (not necessarily disjoint). The Lebesgue integral is then

\int f \, d\mu = \sum_{k=1}^n c_k \mu(A_k),

provided the right-hand side is finite; if any \mu(A_k) = \infty and c_k > 0, or if the sum diverges, the integral equals \infty.^[11] This definition preserves linearity and monotonicity, aligning with the general properties of the Lebesgue integral.^[12] For a general non-negative measurable function f: X \to [0, \infty], the integral is defined as the supremum over all simple functions s with $0 \leq s \leq f:

\int f \, d\mu = \sup \left\{ \int s \, d\mu \mid 0 \leq s \leq f, \, s \text{ simple} \right\}.

The monotone convergence theorem ensures that if \{s_m\} is an increasing sequence of simple functions converging pointwise to f, then \int f \, d\mu = \lim_{m \to \infty} \int s_m \, d\mu.^[11] Since \mu(\{x\}) = 1 for each x \in X, this integral equals \sum_{x \in X} f(x), where the sum is over the (at most countable) support of f and may be \infty. For signed measurable functions f = f^+ - f^- with both f^+ and f^- non-negative, the integral \int f \, d\mu is finite only if \sum_{x \in X} |f(x)| < \infty, in which case it equals \sum_{x \in X} f(x).^[11] As an example, consider the space (\mathbb{N}, 2^{\mathbb{N}}, \mu) with the counting measure \mu and the function f(n) = 1/n^2 for n \in \mathbb{N}. This f is non-negative and measurable, so

\int f \, d\mu = \sum_{n=1}^\infty \frac{1}{n^2} = \frac{\pi^2}{6}.

The series converges absolutely, confirming the integral is finite.^[13]

Relation to Summation

In the context of the Lebesgue integral with respect to the counting measure \mu on a countable set X, the integral of a function f: X \to \mathbb{R} reduces precisely to the corresponding series summation: \int_X f \, d\mu = \sum_{x \in X} f(x), where the sum is effectively over the support of f (points where f(x) \neq 0).^[14] This equivalence holds because the counting measure assigns mass 1 to each singleton \{x\}, so the integral over simple approximations to f yields sums of f(x) weighted by these masses, and the general case follows by monotone convergence for non-negative functions or decomposition into positive and negative parts for signed functions./03%3A_Distributions/3.10%3A_The_Integral_With_Respect_to_a_Measure) For non-negative measurable functions f \geq 0, the integral \int_X f \, d\mu equals the supremum of integrals of simple functions below f, which directly corresponds to the (possibly infinite) sum \sum_{x \in X} f(x).^[14] For signed functions, the integral is defined only when both \int_X f^+ \, d\mu < \infty and \int_X f^- \, d\mu < \infty, where f^+ and f^- are the positive and negative parts; this is equivalent to the absolute convergence of the series \sum_{x \in X} |f(x)|, ensuring the integral equals \sum_{x \in X} f(x). If the series \sum |f(x)| diverges, the integral is infinite, mirroring the divergence of the series. The partial sums of the series \sum f(x) relate to the integral via approximations over finite subsets; for instance, the integral over a finite partial set X_n \subset X is exactly the partial sum \sum_{x \in X_n} f(x), and the full integral is the limit as n \to \infty under monotone convergence.^[14] When the direct series diverges (e.g., oscillates or grows without bound), the Lebesgue integral with counting measure yields infinity, but alternative summation methods such as Cesàro means—defined as the limit of averages of partial sums—can assign finite values to certain divergent series, providing a way to extend summation beyond absolute convergence.^[15] Unlike the Riemann integral, which approximates areas under continuous curves over intervals using limits of sums over partitions, the counting measure integral is inherently discrete and always reduces to an exact (possibly infinite) sum over points, without reliance on partition refinements or continuity assumptions.^[16] For example, consider the harmonic series on the natural numbers with f(n) = 1/n; the integral \int_{\mathbb{N}} (1/n) \, d\mu = \sum_{n=1}^\infty 1/n = \infty, diverging as expected since the partial sums grow logarithmically without bound.^[14]

Applications and Extensions

In Probability and Discrete Spaces

In probability theory, the counting measure serves as the foundational reference measure for discrete probability spaces. On a finite nonempty set X, the normalized counting measure defines the uniform probability measure P, where P(A) = |A| / |X| for any subset A \subseteq X.^[17] This construction yields the discrete uniform distribution, under which every singleton \{x\} \in X has equal probability $1 / |X|, modeling scenarios where all outcomes are equally likely, such as fair dice rolls or random selection from a finite population.^[18] For countable infinite sets X, such as the natural numbers [\mathbb{N}](/page/N+), the counting measure \mu(X) = \infty prevents normalization to a probability measure, as no such P can satisfy P(X) = 1 while assigning positive mass to each point.^[17] However, the unnormalized counting measure functions as an improper prior in Bayesian statistics on discrete parameter spaces, providing a non-informative baseline that integrates to infinity but yields proper posteriors under suitable likelihoods.^[19] It also arises in limiting constructions, such as the intensity measure for Poisson point processes on countable spaces.^[20] The expectation of a random variable X on a discrete probability space (X, \mathcal{P}(X), P), where P is absolutely continuous with respect to the counting measure, is given by the Lebesgue integral E[X] = \int_X x \, dP(x).^[21] With probability density p_x = P(\{x\}), this reduces to the weighted sum E[X] = \sum_{x \in X} x p_x, bridging measure-theoretic integration with classical summation in discrete settings.^[21] A representative example is the geometric distribution on \mathbb{N}, modeling the number of failures before the first success in independent Bernoulli trials with success probability p \in (0,1]. The probability measure assigns P(\{n\}) = (1-p)^n p for n \in \mathbb{N}, defined on the power set \mathcal{P}(\mathbb{N}) with the counting measure as reference, ensuring the space supports countable additivity and total mass 1.^[22] In general, the counting measure on the power set \mathcal{P}(X) of a countable set X equips the space with the discrete \sigma-algebra, where every subset is measurable, facilitating the construction of any discrete probability distribution as a density with respect to this measure.^[23] This structure underpins empirical and combinatorial probability models, ensuring compatibility with measure-theoretic axioms.^[23]

In Functional Analysis

In functional analysis, the counting measure on the natural numbers \mathbb{N} plays a fundamental role in constructing the sequence spaces \ell^p for $1 \leq p \leq \infty, which are precisely the L^p spaces L^p(\mathbb{N}, \mu) where \mu is the counting measure. These spaces consist of all sequences f = (f(n))_{n \in \mathbb{N}} such that \sum_{n=1}^\infty |f(n)|^p < \infty for $1 \leq p < \infty, with the associated p-norm given by

\|f\|_p = \left( \sum_{n=1}^\infty |f(n)|^p \right)^{1/p}.

For p = \infty, \ell^\infty comprises bounded sequences with \|f\|_\infty = \sup_{n \in \mathbb{N}} |f(n)|. This identification arises because integration with respect to the counting measure reduces to summation, endowing \ell^p with the structure of a measure-theoretic function space.^[24]^[25] The space \ell^2(\mathbb{N}), corresponding to L^2(\mathbb{N}, \mu), is a Hilbert space with the inner product

\langle f, g \rangle = \sum_{n=1}^\infty f(n) \overline{g(n)},

which induces the norm \|f\|_2 = \sqrt{\langle f, f \rangle}. This inner product structure facilitates the study of orthogonal bases and spectral theory in discrete settings, mirroring continuous Hilbert spaces like L^2(\mathbb{R}) but adapted to summability conditions. The completeness of \ell^p for $1 \leq p \leq \infty follows directly from the general Riesz-Fischer theorem for L^p spaces, establishing \ell^p as Banach spaces under their norms.^[25]^[24] Operators on these spaces induced by the counting measure include multiplication operators and shift operators. A multiplication operator M_\phi on \ell^p is defined by (M_\phi f)(n) = \phi(n) f(n), where \phi \in \ell^\infty, ensuring boundedness with \|M_\phi\| = \|\phi\|_\infty. Shift operators, such as the unilateral forward shift S f(n) = f(n-1) for n \geq 2 and S f(1) = 0 on \ell^p(\mathbb{N}), or the bilateral shift on \ell^p(\mathbb{Z}), are bounded linear isometries with operator norm 1, preserving the \ell^p norm due to the discrete, translation-invariant nature of the counting measure.^[25]^[26] An illustrative application arises in Fourier analysis on the integers \mathbb{Z} equipped with counting measure, where L^2(\mathbb{Z}, \mu) = \ell^2(\mathbb{Z}), and the Fourier transform yields the discrete Fourier transform, mapping sequences to periodic functions on the unit circle via characters of the group. This framework underpins harmonic analysis on discrete abelian groups, with Plancherel's theorem ensuring unitarity of the transform between \ell^2(\mathbb{Z}) and L^2(\mathbb{T}).^[27]

References

[1]
[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 ...<|control11|><|separator|>
[2]
Measure Theory Basics - UC Berkeley Statistics
Aug 24, 2023 · Example 1 (Counting measure): If X is countable, e.g. X = Z , then a natural measure is the counting measure # ( A ) , which simply counts the ...
[3]
https://stats.libretexts.org/Bookshelves/Probability_Theory/Probability_Mathematical_Statistics_and_Stochastic_Processes_(Siegrist)/01%3A_Foundations/1.07%3A_Counting_Measure
[4]
[PDF] Measures - UC Davis Math
Then ν is a measure, called count- ing measure on X. Every subset of X is measurable with respect to ν. Counting measure is finite if X is finite and σ-finite ...
[5]
Counting measure is the countable sum of the Dirac measures
Oct 19, 2020 · Counting measure is the countable sum of the Dirac measures · 1. Yes it is. Just evaluate the sum of the dirac measures on a set, and see what ...Sum of dirac measure is a measure [closed] - Math Stack ExchangeProof that a measure is a countable sum of dirac measuresMore results from math.stackexchange.com
[6]
[PDF] An Introduction to Measure Theory - Terry Tao
In later sections, we will formally define Lebesgue measure and the Lebesgue ... in measure theory is that one can cut up a single epsilon into count-.
[7]
[PDF] spl34.tex Lecture 4. 17.10.2011. Measures (continued). Completion ...
Oct 17, 2011 · The measure space is purely atomic – has no non- trivial null sets (the only set of cardinality 0 is empty!). Counting measure is useful in ...<|control11|><|separator|>
[8]
Counting Measure - Random Services
The function # on P ( S ) is called counting measure . Counting measure plays a fundamental role in discrete probability structures, and particularly those that ...
[9]
Additive Measures - University of Pittsburgh
The measures of the infinitely many rationals in the interval from 0 to 1 would sum to infinity. This choice would be a quite troublesome, since the rational ...
[10]
[PDF] Chapter 17. General Measure Spaces: Their Properties and ...
Apr 15, 2019 · The counting measure on an uncountable set is not σ-finite. Note. Many of the properties of Lebesgue measure and Lebesgue integration hold.
[11]
[PDF] MEASURE AND INTEGRATION - ETH Zürich
... Lebesgue integral. Definition 1.25 (Step Function). Let X be a set. A ... counting measure. µ : A → [0,∞] is defined by µ(A) := #A for A ∈ A. As an ...
[12]
[PDF] notes on measure theory and the lebesgue integral - People
Mar 12, 2019 · NOTES ON MEASURE THEORY AND THE LEBESGUE INTEGRAL. MAA5229, SPRING 2019 5 is a measure on (X,2X), called counting measure. It is finite if ...
[13]
[PDF] Numerous Proofs of ζ(2) =
First, let's discuss his original “proof” and then fill in some of the gaps with some rigorous analysis afterwards. Theorem 1. ζ(2) = π2. 6. Proof #1, Euler ( ...<|control11|><|separator|>
[14]
[PDF] Lecture 2. Measure and Integration - LSU Math
The sum and product of two measurable functions are measurable functions. (assuming that the sum/product is defined; i.e. we don't have sums like ∞ + (−∞) or.
[15]
245A, Notes 3: Integration on abstract measure spaces ... - Terry Tao
Sep 25, 2010 · On any measure space, one can set up the unsigned and absolutely convergent integrals in almost exactly the same way as was done in the previous ...<|separator|>
[16]
[PDF] MEASURES
A measure is a mathematical object that quantifies the size of sets, like length, area, and volume, which are rules associating a number to sets.
[17]
Discrete Uniform Distributions - Random Services
The discrete uniform distribution is a special case of the general uniform distribution with respect to a measure, in this case counting measure on all subsets ...
[18]
Improper priors and improper posteriors - Wiley Online Library
Jul 14, 2021 · The posterior, however, is in this case improper for and for . There is nothing wrong with observing or , and a ...
[19]
Improper Priors Are Not Improper - jstor
It is well known that improper priors in Bayesian statistics may lead to ... , where is counting measure. The theory presented by Hartigan (1983) does ...
[20]
[PDF] Lp spaces - UC Davis Math
Density theorems enable us to prove properties of Lp functions by proving them for functions in a dense subspace and then extending the result by continuity. ...
[21]
245B, notes 3: L^p spaces
### Summary of Relevant Sections from https://terrytao.wordpress.com/2009/01/09/245b-notes-3-lp-spaces/
[22]
[PDF] arXiv:2107.12103v3 [math.DS] 7 Jun 2022 Shift-like Operators on L (X)
Jun 7, 2022 · Shift-like operators appear naturally as composition operators on Lp(X) when the underlying space is a dissipative measure system. In the ...
[23]
None
### Summary of Harmonic Analysis on Integers ℤ with Counting Measure, Relation to ℓ², and Fourier Transform on Discrete Groups