Measure (mathematics)

In mathematics, the concept of a measure is a generalization and formalization of geometrical measures (length, area, volume) and other common notions, such as magnitude, mass, and probability of events. These seemingly distinct concepts have many similarities and can often be treated together in a single mathematical context. Measures are foundational in probability theory, integration theory, and can be generalized to assume negative values, as with electrical charge. Far-reaching generalizations (such as spectral measures and positive operator-valued measures) of measure are widely used in quantum physics and physics in general.

The intuition behind this concept dates back to Ancient Greece, when Archimedes tried to calculate the area of a circle. But it was not until the late 19th and early 20th centuries that measure theory became a branch of mathematics. The foundations of modern measure theory were laid in the works of Émile Borel, Henri Lebesgue, Nikolai Luzin, Johann Radon, Constantin Carathéodory, and Maurice Fréchet, among others. According to Thomas W. Hawkins Jr., "It was primarily through the theory of multiple integrals and, in particular the work of Camille Jordan that the importance of the notion of measurability was first recognized."

Let \(X\) be a set and \(\Sigma\) a σ-algebra over \(X\), defining subsets of \(X\) that are "measurable". A set function \(\mu\) from \(\Sigma\) to the interval \([0,+\infty]\), that is, the non-negative real number line together with new (so-called infinite) value \(+\infty\), used to denote elements greater than all other (so-called finite) elements, is called a measure if the following conditions hold:

• \(\mu(\varnothing) = 0\)
• Countable additivity (or σ-additivity): For all countable collections \(\{ E_k \}_{k=1}^\infty\) of pairwise disjoint sets in \(\Sigma\),\[\mu{\left(\bigcup_{k=1}^\infty E_k\right)} = \sum_{k=1}^\infty \mu(E_k)\]

If at least one set \(E\) has finite measure, then the requirement \(\mu(\varnothing) = 0\) is met automatically due to countable additivity:\[\mu(E)=\mu(E \cup \varnothing) = \mu(E) + \mu(\varnothing),\]and therefore \(\mu(\varnothing)=0.\)

Note that any sum involving \(+\infty\) will equal \(+\infty\), that is, \(a + \infty = +\infty\) for all \(a\) in the extended reals.

If the condition of non-negativity is dropped, and \(\mu(E)\) only ever equals one of \(+\infty\), \(-\infty\), i.e. no two distinct sets have measures \(+\infty\), \(-\infty\), respectively, then \(\mu\) is called a signed measure.

The pair \((X, \Sigma)\) is called a measurable space, and the members of \(\Sigma\) are called measurable sets.

A triple \((X, \Sigma, \mu)\) is called a measure space. A probability measure is a measure with total measure one – that is, \(\mu(X) = 1.\) A probability space is a measure space with a probability measure.

For measure spaces that are also topological spaces various compatibility conditions can be placed for the measure and the topology. Most measures met in practice in analysis (and in many cases also in probability theory) are Radon measures (usually defined on Hausdorff spaces). When working with locally compact Hausdorff spaces, Radon measures have an alternative, equivalent definition in terms of linear functionals on the locally convex topological vector space of continuous functions with compact support. This approach is taken by Bourbaki (2004) and a number of other sources. For more details, see the article on Radon measures.

Some important measures are listed here.

• The counting measure is defined by \(\mu(S)\) = number of elements in \(S.\)
• The Lebesgue measure on \(\R\) is a complete translation-invariant measure on a σ-algebra containing the intervals in \(\R\) such that \(\mu([0, 1]) = 1\); and every other measure with these properties extends the Lebesgue measure.
• The arc length of an interval on the unit circle in the Euclidean plane extends to a measure on the \(\sigma\)-algebra they generate. It can be called angle measure since the arc length of an interval equals the angle it supports. This measure is invariant under rotations preserving the circle. Similarly, hyperbolic angle measure is invariant under squeeze mapping.
• The Haar measure for a locally compact topological group. For example, \(\mathbb R\) is such a group and its Haar measure is the Lebesgue measure; for the unit circle (seen as a subgroup of the multiplicative group of \(\mathbb C\)) its Haar measure is the angle measure. For a discrete group the counting measure is a Haar measure.
• Every (pseudo) Riemannian manifold \((M,g)\) has a canonical measure \(\mu_g\) that in local coordinates \(x_1,\ldots,x_n\) looks like \(\sqrt{\left|\det g \right|}d^nx\) where \(d^nx\) is the usual Lebesgue measure.
• The Hausdorff measure is a generalization of the Lebesgue measure to sets with non-integer dimension, in particular, fractal sets.
• Every probability space gives rise to a measure which takes the value 1 on the whole space (and therefore takes all its values in the unit interval [0, 1]). Such a measure is called a probability measure or distribution. See the list of probability distributions for instances.
• The Dirac measure δ_a (cf. Dirac delta function) is given by δ_a(S) = χ_S(a), where χ_S is the indicator function of \(S.\) The measure of a set is 1 if it contains the point \(a\) and 0 otherwise.

Other 'named' measures used in various theories include: Borel measure, Jordan measure, ergodic measure, Gaussian measure, Baire measure, Radon measure, Young measure, and Loeb measure.

In physics an example of a measure is spatial distribution of mass (see for example, gravity potential), or another non-negative extensive property, conserved (see conservation law for a list of these) or not. Negative values lead to signed measures, see "generalizations" below.

• Liouville measure, known also as the natural volume form on a symplectic manifold, is useful in classical statistical and Hamiltonian mechanics.
• Gibbs measure is widely used in statistical mechanics, often under the name canonical ensemble.

A measurable set \(X\) is called a null set if \(\mu(X) = 0.\) A subset of a null set is called a negligible set. A negligible set need not be measurable, but every measurable negligible set is automatically a null set. A measure is called complete if every negligible set is measurable.

A measure can be extended to a complete one by considering the σ-algebra of subsets \(Y\) which differ by a negligible set from a measurable set \(X,\) that is, such that the symmetric difference of \(X\) and \(Y\) is contained in a null set. One defines \(\mu(Y)\) to equal \(\mu(X).\)

If \(f:X\to[0,+\infty]\) is \((\Sigma,{\cal B}([0,+\infty]))\)-measurable, then \[\mu\{x\in X: f(x) \geq t\} = \mu\{x\in X: f(x) > t\}\] for almost all \(t \in [-\infty,\infty].\) This property is used in connection with Lebesgue integral.

Both \(F(t) := \mu\{x\in X : f(x) > t\}\) and \(G(t) := \mu\{x\in X : f(x) \geq t\}\) are monotonically non-increasing functions of \(t,\) so both of them have at most countably many discontinuities and thus they are continuous almost everywhere, relative to the Lebesgue measure. If \(t < 0\) then \(\{x\in X : f(x) \geq t\} = X = \{x\in X : f(x) > t\},\) so that \(F(t) = G(t),\) as desired.

If \(t\) is such that \(\mu\{x\in X : f(x) > t\} = +\infty\) then monotonicity implies \[\mu\{x\in X : f(x) \geq t\} = +\infty,\] so that \(F(t) = G(t),\) as required. If \(\mu\{x\in X : f(x) > t\} = +\infty\) for all \(t\) then we are done, so assume otherwise. Then there is a unique \(t_0 \in \{-\infty\} \cup [0,+\infty)\) such that \(F\) is infinite to the left of \(t\) (which can only happen when \(t_0 \geq 0\)) and finite to the right. Arguing as above, \(\mu\{x\in X : f(x) \geq t\} = +\infty\) when t < t_0.\) Similarly, if \(t_0 \geq 0\) and \(F\left(t_0\right) = +\infty\) then \(F\left(t_0\right) = G\left(t_0\right).\)

For \(t > t_0,\) let \(t_n\) be a monotonically non-decreasing sequence converging to \(t.\) The monotonically non-increasing sequences \(\{x\in X : f(x) > t_n\}\) of members of \(\Sigma\) has at least one finitely \(\mu\)-measurable component, and \[\{x\in X : f(x) \geq t\} = \bigcap_n \{x\in X : f(x) > t_n\}.\] Continuity from above guarantees that \[\mu\{x\in X : f(x) \geq t\} = \lim_{t_n \uparrow t} \mu\{x\in X : f(x) > t_n\}.\] The right-hand side \(\lim_{t_n \uparrow t} F\left(t_n\right)\) then equals \(F(t) = \mu\{x\in X : f(x) > t\}\) if \(t\) is a point of continuity of \(F.\) Since \(F\) is continuous almost everywhere, this completes the proof.

Measures are required to be countably additive. However, the condition can be strengthened as follows. For any set \(I\) and any set of nonnegative \(r_i\) where \(i\in I\) define: \[\sum_{i\in I} r_i=\sup\left\lbrace\sum_{i\in J} r_i : |J|<\infty, J\subseteq I\right\rbrace.\] That is, we define the sum of the \(r_i\) to be the supremum of all the sums of finitely many of them.

A measure \(\mu\) on \(\Sigma\) is \(\kappa\)-additive if for any \(\lambda<\kappa\) and any family of disjoint sets \(X_\alpha,\alpha<\lambda\) the following hold: \[\bigcup_{\alpha\in\lambda} X_\alpha \in \Sigma\] \[\mu\left(\bigcup_{\alpha\in\lambda} X_\alpha\right) = \sum_{\alpha\in\lambda}\mu\left(X_\alpha\right).\] The second condition is equivalent to the statement that the ideal of null sets is \(\kappa\)-complete.

A measure space \((X, \Sigma, \mu)\) is called finite if \(\mu(X)\) is a finite real number (rather than \(\infty\)). A measure \(\mu\) is called σ-finite if \(X\) can be decomposed into a countable union of measurable sets of finite measure.

For example, the real numbers with the standard Lebesgue measure are σ-finite but not finite. The natural numbers are also σ-finite with respect to counting measure. However, the real numbers are not σ-finite with respect to counting measure.

Let \(X\) be a set, let \({\cal A}\) be a sigma-algebra on \(X,\) and let \(\mu\) be a measure on \({\cal A}.\) We say \(\mu\) is semifinite to mean that for all \(A\in\mu^\text{pre}\{+\infty\},\) \({\cal P}(A)\cap\mu^\text{pre}(\R_{>0})\ne\emptyset.\)

Semifinite measures generalize sigma-finite measures, in such a way that some big theorems of measure theory that hold for sigma-finite but not arbitrary measures can be extended with little modification to hold for semifinite measures. (To-do: add examples of such theorems; cf. the talk page.)

Localizable measures are a special case of semifinite measures and a generalization of sigma-finite measures.

Let \(X\) be a set, let \({\cal A}\) be a sigma-algebra on \(X,\) and let \(\mu\) be a measure on \({\cal A}.\)

• Let \(\mathbb F\) be \(\R\) or \(\C,\) and let \(T : L_\mathbb{F}^\infty(\mu) \to \left(L_\mathbb{F}^1(\mu)\right)^* : g \mapsto T_g = \left(\int fgd\mu\right)_{f\in L_\mathbb{F}^1(\mu)}.\) Then \(\mu\) is localizable if and only if \(T\) is bijective (if and only if \(L_\mathbb{F}^\infty(\mu)\) "is" \(L_\mathbb{F}^1(\mu)^*\)).

A measure is said to be s-finite if it is a countable sum of finite measures. S-finite measures are more general than sigma-finite ones and have applications in the theory of stochastic processes.

If the axiom of choice is assumed to be true, it can be proved that not all subsets of Euclidean space are Lebesgue measurable; examples of such sets include the Vitali set, and the non-measurable sets postulated by the Hausdorff paradox and the Banach-Tarski paradox.

For certain purposes, it is useful to have a "measure" whose values are not restricted to the non-negative reals or infinity. For instance, a countably additive set function with values in the (signed) real numbers is called a signed measure, while such a function with values in the complex numbers is called a complex measure. Observe, however, that complex measure is necessarily of finite variation, hence complex measures include finite signed measures but not, for example, the Lebesgue measure.

Measures that take values in Banach spaces have been studied extensively. A measure that takes values in the set of self-adjoint projections on a Hilbert space is called a projection-valued measure; these are used in functional analysis for the spectral theorem. When it is necessary to distinguish the usual measures which take non-negative values from generalizations, the term positive measure is used. Positive measures are closed under conical combination but not general linear combination, while signed measures are the linear closure of positive measures. More generally see measure theory in topological vector spaces.

Another generalization is the finitely additive measure, also known as a content. This is the same as a measure except that instead of requiring countable additivity we require only finite additivity. Historically, this definition was used first. It turns out that in general, finitely additive measures are connected with notions such as Banach limits, the dual of \(L^\infty\) and the Stone-Čech compactification. All these are linked in one way or another to the axiom of choice. Contents remain useful in certain technical problems in geometric measure theory; this is the theory of Banach measures.

A charge is a generalization in both directions: it is a finitely additive, signed measure. (Cf. ba space for information about bounded charges, where we say a charge is bounded to mean its range is a bounded subset of R.)