1 Foundations
Measure theory arose from the need to make precise the intuitive ideas of size, area, and probability. Classical geometry handled simple shapes well, but analysis required a framework that could treat irregular sets, limits of functions, and infinite processes in a consistent way. The subject supplies that framework and has become central to modern analysis.
1.1 Historical background
Early ideas of measure can be traced to ancient geometry, where area and volume were computed for familiar figures. In the nineteenth century, attempts to rigorously define length and area led to deeper work by mathematicians such as Borel, Lebesgue, and others. Their efforts replaced ad hoc geometric arguments with abstract set-based definitions.
Lebesgue’s work on integration was especially influential. It showed that many functions not suitable for classical Riemann integration could still be integrated in a natural way. This opened the door to new results in analysis and probability.
1.2 Motivating problems in analysis
Several problems motivated the development of measure theory. One was the need to integrate limits of functions without losing control over convergence. Another was the desire to assign a consistent notion of size to complicated sets, including sets formed by countable unions and intersections.
A further motivation came from probability, where “size” must represent likelihood rather than geometric extent. Measure theory provided a common language for both geometry and randomness, allowing probabilistic ideas to be studied with analytic tools.
1.3 Basic set theory and set operations
Measure theory is built on sets and operations on sets. The basic objects are collections of points, and the fundamental relations are inclusion, equality, and membership. From these, one forms larger families of sets that are stable under natural operations.
1.3.1 Union, intersection, and complement
The union of sets collects all elements belonging to at least one set, while the intersection contains elements shared by all of them. The complement of a set consists of all points outside it, relative to a fixed ambient space.
These operations are essential because measure theory often studies how size behaves under decomposition and recombination. A consistent notion of measure must interact well with unions of disjoint sets and with complements inside a larger space.
1.3.2 Families of sets
A family of sets is a collection chosen for a shared purpose, often closed under certain operations. In measure theory, the most important families are those stable under countable unions, countable intersections, and complements.
Such families allow one to define measurable sets and to ensure that limits of set operations remain within the same framework. This stability is crucial for building a theory that supports integration and convergence.
2 Measures and measurable spaces
The central concepts of the subject are measurable spaces and measures. A measurable space specifies which sets are eligible for measurement, and a measure assigns a nonnegative size to those sets in a way that respects countable additivity. Together they form the basic setting for integration and probability.
2.1 Sigma-algebras
A sigma-algebra is a collection of sets closed under complements and countable unions. It determines which sets are regarded as measurable within a given space.
2.1.1 Definitions and examples
Typical examples include the collection of all subsets of a finite set, the Borel sets on the real line, and the collection generated by open sets in a topological space. These examples show that sigma-algebras can be either very small or very large, depending on the application.
The closure properties guarantee that many natural limiting operations remain measurable. This makes sigma-algebras well suited to analysis, where countable processes are ubiquitous.
2.1.2 Generated sigma-algebras
Given a set of subsets, one can form the smallest sigma-algebra containing them. This generated sigma-algebra includes all sets obtainable by repeatedly applying countable unions, intersections, and complements.
This construction is important because one often begins with a simple collection, such as open intervals, and then extends it to a richer measurable structure. In this way, a broad theory can be built from a modest starting point.
2.2 Measurable spaces
A measurable space consists of a set together with a chosen sigma-algebra on that set. The sigma-algebra identifies the measurable subsets, while the underlying set serves as the ambient universe.
Measurable spaces provide the stage on which measures are defined. Different sigma-algebras on the same set can lead to very different theories, depending on how much detail is needed.
2.3 Measures
A measure is a function that assigns a nonnegative extended real number to measurable sets. It must satisfy countable additivity on disjoint collections, meaning the measure of a union equals the sum of the measures.
This property captures the idea that size should be compatible with decomposition. If a set is split into disjoint pieces, its total measure should equal the combined measure of those pieces.
2.3.1 Finitely additive and countably additive set functions
A finitely additive set function preserves addition for finite disjoint unions. Countable additivity is stronger and is the key requirement in measure theory.
The stronger condition is essential for handling limits and infinite decompositions. It ensures that the measure behaves well under the kinds of limiting arguments that occur throughout analysis.
2.3.2 Examples of measures
Many familiar notions of size are examples of measures. These include counting, length, volume, and probability.
2.3.2.1 Counting measure
Counting measure assigns to a set the number of elements it contains, with infinite values allowed for infinite sets. It is especially useful on discrete spaces and in combinatorial contexts.
2.3.2.2 Lebesgue measure
Lebesgue measure is the standard notion of length, area, and volume on Euclidean spaces. It agrees with intuitive geometric size on intervals and rectangles and extends this notion to far more complicated sets.
2.3.2.3 Probability measures
A probability measure assigns total mass one to the entire space. It interprets the measure of a set as the likelihood that an outcome lies in that set.
2.4 Outer measures
An outer measure assigns size to all subsets of a given set, not only to measurable ones. It is typically easier to define than a full measure and serves as a starting point for construction.
Outer measures are especially useful in building measures from geometric approximations. They provide a flexible tool for extending size concepts beyond initially simple families of sets.
2.4.1 Carathéodory measurability
Carathéodory measurability identifies those sets that interact cleanly with an outer measure. A set is measurable if splitting any other set along it does not distort the total outer measure.
This criterion selects a sigma-algebra on which the outer measure becomes an actual measure. It is one of the most elegant abstract mechanisms in the subject.
2.4.2 Construction of measures from outer measures
A common strategy is to define an outer measure first and then isolate the measurable sets by Carathéodory’s criterion. The result is a measure on a sigma-algebra generated by the measurable sets.
This approach is used in the construction of Lebesgue measure and in other settings where direct specification of a measure on all sets is difficult.
3 Construction of classical measures
Classical measures arise from geometric intuition and are extended through abstraction. The standard example is Lebesgue measure, which captures length, area, and volume in Euclidean spaces.
3.1 Lebesgue measure on the real line
On the real line, Lebesgue measure formalizes the idea that intervals have length equal to the difference of their endpoints. From this starting point, the measure is extended to a much larger class of sets.
3.1.1 Length of intervals
For an interval, its measure is its geometric length. This choice is natural and serves as the normalization from which the theory develops.
The definition agrees with ordinary intuition for open, closed, and half-open intervals, with endpoints contributing no length.
3.1.2 Translation invariance
Lebesgue measure is translation invariant: shifting a set along the line does not change its measure. This property reflects the idea that size should not depend on position.
Translation invariance is one of the main reasons Lebesgue measure is regarded as the canonical measure on the real line.
3.1.3 Extension to Borel and Lebesgue sets
Starting from intervals, one can extend measure to Borel sets, which are generated by open sets. A further completion yields Lebesgue measurable sets, a larger class that includes subsets of null sets.
This extension greatly enlarges the scope of analysis, allowing highly irregular sets to be measured while preserving the basic geometric meaning of length.
3.2 Higher-dimensional Lebesgue measure
Lebesgue measure extends naturally to Euclidean spaces of any finite dimension. In these spaces, it measures area in two dimensions, volume in three, and higher-dimensional analogues in general.
3.2.1 Rectangles and boxes
The basic building blocks are rectangles or boxes, whose measure is given by the product of side lengths. More complicated sets are then approximated by combinations of such boxes.
This multiplicative rule matches geometric intuition and supports the construction of measure in several dimensions.
3.2.2 Volume in Euclidean space
Higher-dimensional Lebesgue measure captures the notion of volume in \(\mathbb{R}^n\). It is invariant under translations and, in appropriate form, under rotations as well.
This measure plays a central role in multivariable calculus, partial differential equations, and geometric analysis.
3.3 Completion of measures
A measure is completed by adding all subsets of null sets to the measurable family. This ensures that sets differing only by negligible pieces are treated consistently.
3.3.1 Null sets
A null set has measure zero. Such sets are negligible for many analytic purposes because they do not affect integrals or almost everywhere statements.
3.3.2 Completed sigma-algebras
The completed sigma-algebra includes every subset of every null set. Completion eliminates technical gaps and makes the measurable structure more robust.
4 Measurable functions and sets
Once a measurable space and measure are in place, one studies functions that are compatible with this structure. Measurable functions are the natural objects of integration, and measurable sets serve as their level-set building blocks.
4.1 Measurable sets
A measurable set is one that belongs to the chosen sigma-algebra. Such sets are precisely those for which the measure is defined.
In practice, measurability allows one to combine sets using countable operations without leaving the framework. This closure is fundamental to modern analysis.
4.2 Measurable functions
A function is measurable if the preimage of every open set, or equivalently of every suitable generating set, is measurable. This definition ensures that the function interacts properly with the underlying measurable structure.
Measurable functions include continuous functions, step functions, and many limits of such functions. They form the broad class of functions that can be integrated in the Lebesgue sense.
4.2.1 Equivalent characterizations
Measurability can be characterized in several equivalent ways, using preimages of open intervals, upper or lower level sets, or approximation by simple functions. These equivalences make the concept flexible and easy to use.
Such characterizations are often chosen according to the task at hand. For example, level-set descriptions are useful in proofs involving inequalities, while approximation by simple functions is central to integration.
4.2.2 Operations preserving measurability
Sums, products, scalar multiples, and pointwise limits of measurable functions are measurable under suitable conditions. Composition with continuous functions also preserves measurability.
These closure properties allow complex expressions to be built from simple measurable pieces without leaving the theory.
4.3 Simple functions
A simple function takes only finitely many values, each on a measurable set. These functions are the basic approximants used to construct the integral.
Because they are easy to handle, simple functions serve as a bridge between set theory and more general measurable functions.
4.4 Almost everywhere concepts
Many measure-theoretic results are formulated not everywhere, but almost everywhere. This means that a property may fail on a set of measure zero without affecting the conclusion.
4.4.1 Equality almost everywhere
Two functions are equal almost everywhere if they differ only on a null set. In many contexts, such functions are treated as essentially the same.
This viewpoint is especially important in integration and in \(L^p\) spaces, where null-set differences do not change the norm or integral.
4.4.2 Support and essential properties
The support of a function describes where it is nonzero or significantly active, while essential properties ignore changes on null sets. These ideas help capture the effective behavior of measurable functions.
They are used frequently in analysis to state results in a way that is invariant under negligible modifications.
5 Integration theory
Integration theory is one of the main achievements of measure theory. It extends the idea of area under a curve to much broader settings and handles functions with intricate behavior.
5.1 The integral of simple functions
The integral of a simple function is defined by summing the values it takes, each weighted by the measure of the set where it occurs. This is the starting point for the general theory.
Because simple functions approximate more complicated ones, their integrals provide a natural foundation for defining the integral of measurable functions.
5.2 The Lebesgue integral
The Lebesgue integral is defined by approximation from below or by decomposition into positive and negative parts. It allows one to integrate functions that may be too irregular for classical methods.
5.2.1 Positive and signed functions
Nonnegative functions are integrated first, using increasing simple-function approximations. Signed functions are then handled by separating them into positive and negative parts, provided the resulting quantities are well behaved.
This approach gives a systematic and flexible method for extending integration beyond elementary calculus.
5.2.2 Integrability criteria
A function is integrable when its positive and negative parts are both finite in integral. Equivalent criteria are often given in terms of absolute integrability or membership in \(L^1\).
These conditions ensure that the integral is finite and that many limit theorems can be applied.
5.3 Comparison with Riemann integration
Lebesgue integration generalizes Riemann integration but is more powerful in handling limits and discontinuities. Every Riemann-integrable function on a compact interval is Lebesgue integrable, though the converse need not hold.
The Lebesgue approach is especially effective when pointwise behavior is complicated but the size of exceptional sets is small.
5.4 Change of variables
Change of variables formulas describe how integrals transform under coordinate substitutions. They are essential in calculus, geometry, and multivariable analysis.
5.4.1 Substitution formula
In one dimension, substitution replaces integration in one variable by integration in another variable after a suitable transformation. The formula accounts for the stretching or compression caused by the change of coordinates.
5.4.2 Jacobians
In higher dimensions, the Jacobian determinant measures local volume distortion under a transformation. It appears as the correction factor in the multidimensional change of variables formula.
This determinant is central to integrating over curved or transformed regions in Euclidean space.
5.5 Tonelli’s and Fubini’s theorems
Tonelli’s theorem permits the interchange of integrals for nonnegative functions, while Fubini’s theorem extends this to integrable functions of several variables. These results justify repeated integration and the use of iterated integrals.
They are indispensable in analysis and applications, especially when working with product spaces.
6 Convergence theorems
Measure theory provides powerful tools for passing limits through integrals. These convergence theorems are among its most useful results, since they make approximation arguments rigorous.
6.1 Monotone convergence theorem
The monotone convergence theorem states that an increasing sequence of nonnegative measurable functions has integrals converging to the integral of the pointwise limit. It is a cornerstone of the theory.
This result allows one to build integrals from approximations and to justify limiting processes step by step.
6.2 Fatou’s lemma
Fatou’s lemma gives an inequality relating the integral of a pointwise limit inferior to the limit inferior of the integrals. It is often used when exact convergence of integrals is unavailable.
The lemma is a basic tool for handling sequences of nonnegative functions and for proving stronger convergence theorems.
6.3 Dominated convergence theorem
The dominated convergence theorem states that pointwise convergence, together with domination by an integrable function, implies convergence of integrals. It is one of the most widely used results in analysis.
This theorem often applies when a sequence is bounded by a single integrable control function, making it possible to exchange limit and integral safely.
6.4 Convergence in measure
Convergence in measure is weaker than pointwise convergence and describes the idea that functions become close except on sets of small measure. It plays an important role in probabilistic and analytic limits.
6.4.1 Almost everywhere convergence
Almost everywhere convergence means pointwise convergence fails only on a null set. It is a stronger notion than convergence in measure, though the two are related by several classical theorems.
6.4.2 Uniform integrability
Uniform integrability is a condition that prevents mass from escaping into small sets. It is especially useful in studying convergence of integrals and sequences of random variables.
7 Product measures and integration on product spaces
Product spaces allow one to combine measurable spaces and study functions of several variables. The corresponding product measures provide the measure-theoretic basis for multivariable integration.
7.1 Product sigma-algebras
The product sigma-algebra is generated by measurable rectangles in a Cartesian product of spaces. It supplies the measurable sets for several variables at once.
This construction is natural for studying functions depending on two or more coordinates.
7.2 Construction of product measures
A product measure is built from measures on each factor space. On rectangles, its value is determined by multiplying the measures of the factors, and then it is extended to the product sigma-algebra.
This construction is fundamental in probability, where independent random variables are modeled on product spaces.
7.3 Iterated integrals
Iterated integrals integrate a function one variable at a time. Under suitable hypotheses, the order of integration can be exchanged without changing the result.
This technique simplifies many calculations and underlies numerous proofs in analysis.
7.4 Sections and slices of measurable sets
A section or slice of a set in a product space is obtained by fixing some coordinates and varying the others. Measurability of slices is often used to analyze product sets and product functions.
Such sectional arguments are central in proving results about iterated integration and measurable selection.
8 Signed measures and decomposition
Not all set functions of interest are nonnegative. Signed and complex measures extend the theory to objects that can take positive and negative values or complex values.
8.1 Signed and complex measures
A signed measure assigns values that may be positive or negative, while a complex measure takes complex values. These generalizations are useful in harmonic analysis and functional analysis.
They can often be studied through their total variation, which measures overall magnitude.
8.2 Hahn decomposition theorem
The Hahn decomposition theorem states that a space can be split into a positive part and a negative part for a signed measure. On one part the measure behaves nonnegatively, and on the other nonpositively.
This decomposition is a foundational structural result for signed measures.
8.3 Jordan decomposition theorem
The Jordan decomposition theorem expresses a signed measure as the difference of two mutually singular positive measures. This representation is unique and clarifies the structure of signed measures.
It allows one to reduce many questions about signed measures to questions about ordinary measures.
8.4 Absolute continuity and singularity
One measure is absolutely continuous with respect to another if it gives zero to every set negligible for the other. Singular measures are supported on disjoint measurable structures in a strong sense.
These relations describe how measures overlap or separate from one another.
8.4.1 Radon–Nikodym theorem
The Radon–Nikodym theorem states that an absolutely continuous measure has a density with respect to a reference measure. This density is a measurable function that represents the measure by integration.
The theorem is one of the most important bridges between measures and functions.
8.4.2 Lebesgue decomposition theorem
The Lebesgue decomposition theorem splits a measure into an absolutely continuous part and a singular part relative to another measure. This decomposition is useful in analysis, probability, and partial differential equations.
9 Specialized classes of measures
Beyond the general theory, many important measures have additional structure. These special classes are designed to capture geometric or analytic properties needed in applications.
9.1 Regular measures
Regular measures can be approximated well by simpler sets such as open or compact sets. Regularity provides a close connection between abstract measure and topology.
9.1.1 Inner and outer regularity
Inner regularity means a set’s measure can be approximated from within by compact sets. Outer regularity means it can be approximated from outside by open sets.
Together, these properties make measure more tractable on topological spaces.
9.2 Borel measures
Borel measures are defined on the Borel sigma-algebra generated by open sets. They are the natural measures associated with topological spaces.
These measures are often the starting point for more refined constructions such as Radon measures or completed measures.
9.3 Radon measures
Radon measures are regular Borel measures with good finiteness properties on compact sets. They arise naturally in analysis on locally compact spaces.
Their combination of topological and measure-theoretic regularity makes them especially useful in abstract analysis.
9.4 Atomic and non-atomic measures
An atomic measure concentrates positive mass on individual points or minimal measurable pieces. A non-atomic measure has no such point masses, so every set of positive measure can be subdivided further.
This distinction matters in probability, geometry, and ergodic theory.
9.5 Probability measures
Probability measures are normalized measures with total mass one. They form the measure-theoretic foundation of probability spaces and stochastic modeling.
Their axioms align with both intuition and rigorous analysis of random phenomena.
10 Measure theory in related fields
Measure theory influences many branches of mathematics by providing a common language for size, averaging, and limiting behavior. Its methods are especially prominent in probability, analysis, and dynamics.
10.1 Probability theory
Probability theory is built on measure spaces in which the total measure is one. Events are measurable sets, and random variables are measurable functions.
10.1.1 Random variables and expectation
A random variable is a measurable function from a probability space to the real numbers or another measurable space. Its expectation is the integral of the variable with respect to the probability measure.
This framework unifies discrete and continuous probability distributions.
10.1.2 Conditional expectation
Conditional expectation is a measure-theoretic averaging operation relative to partial information. It produces the best integrable approximation within a sub-sigma-algebra.
It is central to martingale theory and modern probability.
10.2 Functional analysis
Measure theory underlies many constructions in functional analysis, particularly spaces of integrable functions and operators defined by integration.
10.2.1 Lp spaces
The \(L^p\) spaces consist of measurable functions whose absolute values have finite \(p\)-th power integral, up to equality almost everywhere. These spaces are fundamental examples of normed and Banach spaces.
They provide a natural setting for analysis of functions by size and integrability.
10.2.2 Duality and norms
Measure-theoretic integration gives rise to duality pairings between function spaces, especially between \(L^p\) and \(L^q\) spaces. Norms in these spaces quantify size in ways adapted to the integral.
This structure is essential in optimization, PDEs, and operator theory.
10.3 Harmonic analysis
Harmonic analysis studies functions through waves, Fourier transforms, and related decompositions. Measure theory supplies the integration framework needed to define and analyze these transforms.
Many results in harmonic analysis depend on measurable functions, product measures, and convergence theorems.
10.4 Ergodic theory
Ergodic theory examines long-term behavior of measure-preserving transformations. It connects dynamical systems with invariant measures and averages over time.
The subject relies heavily on almost everywhere statements and on the interaction between dynamics and measure.
10.5 Geometric measure theory
Geometric measure theory studies geometric objects using measure-theoretic methods. It investigates irregular surfaces, rectifiability, and dimension through measures adapted to geometry.
This field extends classical notions of area and perimeter to far more general sets.
11 Further topics
Advanced measure theory includes refined results about differentiation, exceptional sets, and geometric notions of size. These topics deepen the connection between measure and analysis.
11.1 Differentiation of measures
Differentiation theorems describe how measures can be recovered from local averages over shrinking neighborhoods. They generalize the idea that a function can be obtained from its averages at small scales.
These results are closely related to pointwise behavior of integrable functions.
11.2 Vitali covering theorem
The Vitali covering theorem gives a way to select disjoint or nearly disjoint subsets from a covering family. It is a key tool in differentiation theory and in proving maximal-function estimates.
11.3 Hausdorff measure and dimension
Hausdorff measure generalizes length, area, and volume to sets of fractional dimension. It is used to define Hausdorff dimension, which measures the complexity of fractal-like sets.
This theory links geometric size with scaling behavior.
11.4 Non-measurable sets
Non-measurable sets are subsets for which no consistent extension of a given measure can be made while preserving all desired properties. Their existence shows that not every set can be assigned a well-behaved size.
They are an important reminder that measurability is a substantive restriction, not a trivial one.
11.5 Measure-preserving transformations
A measure-preserving transformation leaves measure unchanged under mapping. Such transformations are central in ergodic theory and dynamical systems.
They formalize the idea of motion or rearrangement without distortion of total size.
12 Applications
Measure theory has broad applications across mathematics and the sciences. Its ability to handle size, randomness, and limiting processes makes it indispensable in both pure and applied settings.
12.1 Statistical modeling
In statistics, measures provide the mathematical basis for distributions, likelihoods, and expectation-based inference. Many statistical models are expressed in terms of probability measures on parameter or sample spaces.
This viewpoint supports rigorous treatment of continuous data and uncertainty.
12.2 Partial differential equations
Measure theory is used in the study of weak solutions, variational methods, and energy estimates. Integrals of functions and their derivatives are often interpreted in the measure-theoretic sense.
This approach is essential when classical smooth solutions do not exist.
12.3 Dynamical systems
In dynamical systems, measures help describe invariant sets, long-term averages, and chaotic behavior. Measure-preserving maps and ergodic averages are key concepts in the field.
The theory provides tools for analyzing systems that evolve over time without regular geometric structure.
12.4 Economics and finance
Measure-theoretic probability is widely used in economics and finance to model uncertainty, random returns, and expected utility. Stochastic processes and pricing methods are typically formulated on probability spaces.
The framework supports precise reasoning about risk, distributions, and temporal dependence.