1 Statement of the theorem

The Lebesgue decomposition theorem describes how one measure can be split relative to another. If two sigma-finite measures are defined on the same measurable space, then one of them can be written as the sum of two uniquely determined parts: one that is absolutely continuous with respect to the other, and one that is singular with respect to it. This result gives a precise way to compare measures by separating the portion that is controlled by a reference measure from the portion that lives on a disjoint null set.

1.1 Measures and sigma-finiteness

The theorem applies to measures on a common measurable space. Sigma-finiteness is a standard regularity condition requiring the space to be covered by countably many measurable sets of finite measure. This hypothesis is important because it ensures that the decomposition behaves well and that related results, especially the Radon–Nikodym theorem, can be used effectively.

1.2 Absolute continuity

A measure is absolutely continuous with respect to another measure if every set that is null for the reference measure is also null for the first measure. In this case, the first measure cannot assign mass to any set invisible to the second. Absolute continuity is the measure-theoretic analogue of being described by a density.

1.3 Singular measures

Two measures are singular when they concentrate on disjoint measurable sets. More precisely, a measure is singular with respect to another if there exists a measurable set on which one measure is supported and the other assigns measure zero. Singular parts capture mass that cannot be represented by a density relative to the reference measure.

1.4 Decomposition into continuous and singular parts

The theorem states that a measure can be decomposed into an absolutely continuous component and a singular component relative to another measure. The absolutely continuous part reflects the overlap with the reference measure, while the singular part accounts for mass concentrated away from it. In many familiar settings, this corresponds to splitting a distribution into a density part and a discrete or otherwise exceptional part.

1.5 Uniqueness of the decomposition

The two parts in the decomposition are uniquely determined. If a measure is written in two ways as the sum of an absolutely continuous part and a singular part relative to the same reference measure, then the corresponding components must agree. Uniqueness makes the theorem a canonical structural result rather than merely an existence statement.

2 Background and prerequisites

Understanding the theorem requires basic measure-theoretic language. The main ingredients are measurable spaces, sigma-algebras, measures, null sets, and the finiteness conditions that allow decomposition arguments to work cleanly. These concepts form the framework in which absolute continuity and singularity are defined.

2.1 Measurable spaces

A measurable space consists of a set together with a collection of subsets designated as measurable. These measurable sets are the domain on which measures are defined. The decomposition theorem concerns how one measure compares with another on the same measurable structure.

2.2 Sigma-algebras and measures

A sigma-algebra is a family of sets closed under complements and countable unions. Measures assign nonnegative extended real values to sets in a sigma-algebra in a way that is countably additive. This additivity is essential for decomposing a measure into meaningful components.

2.3 Null sets and support concepts

Null sets are sets of measure zero. They are central because absolute continuity is defined in terms of preservation of null sets. Support, while sometimes used informally, indicates where a measure concentrates most of its mass and helps visualize the singular and continuous parts of a decomposition.

2.4 Sigma-finite measure spaces

A measure space is sigma-finite when it can be covered by countably many sets of finite measure. This condition is weaker than finiteness but still strong enough for many structural theorems. In decomposition theory, sigma-finiteness prevents pathological behavior and permits countable constructions.

3 Relation to the Radon–Nikodym theorem

The Lebesgue decomposition theorem is closely linked to the Radon–Nikodym theorem. The latter provides a derivative, or density, for an absolutely continuous measure with respect to a reference measure. The former first separates out the singular part, then identifies the density of the remaining component.

3.1 Deriving the absolutely continuous part

Once the singular portion is removed, the remaining measure is absolutely continuous with respect to the reference measure. The Radon–Nikodym theorem then guarantees a measurable density function representing that part. Thus the decomposition theorem creates the setting in which the derivative theorem can be applied.

3.2 Radon–Nikodym derivative

The Radon–Nikodym derivative is the measurable function whose integral over a set reproduces the absolutely continuous measure of that set. It generalizes the idea of a density in calculus. In the context of Lebesgue decomposition, it describes the continuous component of the measure relative to the chosen reference measure.

3.3 Comparison with the Lebesgue decomposition theorem

The Radon–Nikodym theorem addresses only the absolutely continuous case, while the Lebesgue decomposition theorem handles the full measure by splitting off the singular component first. Together, the two theorems give a complete picture of how one measure can be expressed relative to another.

3.3.1 Special case of finite measures

For finite measures, the decomposition theorem takes a simpler form because the total mass is bounded. Many proofs and examples are easiest in this setting, and the connection with densities is especially transparent.

3.3.2 Role of sigma-finiteness

Sigma-finiteness is the bridge between the decomposition theorem and the Radon–Nikodym theorem. Without it, the existence of a density for the absolutely continuous part may fail, and the decomposition may not be available in the standard form. This hypothesis is therefore a key technical assumption rather than a mere convenience.

4 Proof of the theorem

Standard proofs use maximality arguments and the Radon–Nikodym theorem. The general strategy is to identify the largest absolutely continuous measure dominated by the given measure, subtract it, and show that the remainder must be singular. The uniqueness then follows from the way these parts are characterized.

4.1 Construction of the absolutely continuous component

One begins by considering measures absolutely continuous with respect to the reference measure and dominated by the original measure. A maximal such measure can be obtained using order-theoretic arguments. This maximal part becomes the absolutely continuous component of the decomposition.

4.2 Construction of the singular component

After the absolutely continuous portion is identified, the difference between the original measure and this part is shown to be singular. The singularity follows because any overlap with the reference measure would contradict maximality. This step establishes that the remainder is concentrated on a set invisible to the reference measure.

4.3 Use of maximality arguments

Maximality is often proved using a comparison of candidate measures and the fact that sums of absolutely continuous measures remain absolutely continuous. By enlarging a candidate whenever possible, one obtains an extremal measure with the desired property. Such arguments are common in measure theory because they turn existence into a problem of order and domination.

4.4 Uniqueness proof

Uniqueness is established by comparing two possible decompositions and observing that the differences between corresponding parts would have to be both absolutely continuous and singular. The only measure with both properties is the zero measure. This forces the two decompositions to coincide.

4.5 Alternative proof strategies

Alternative proofs may use functional-analytic methods, such as the Hahn-Banach theorem, or rely more directly on the Radon–Nikodym theorem in finite settings. Some presentations proceed through signed measures or through the structure of measures on complete spaces. These approaches differ in style but lead to the same canonical decomposition.

5 Variants and generalizations

The core idea of decomposition extends beyond nonnegative measures. Signed measures, complex measures, and vector-valued measures all admit analogous structural results under appropriate assumptions. These generalizations broaden the theorem’s reach in analysis.

5.1 Decomposition of signed measures

For signed measures, one typically combines the Lebesgue decomposition with the Jordan decomposition. The measure is split into positive and negative parts, and each part can then be decomposed relative to a reference measure. This yields a refined structural description of signed mass.

5.2 Decomposition of complex measures

Complex measures can also be decomposed into absolutely continuous and singular components. Since complex measures may oscillate, their analysis often proceeds through variation measures. The decomposition is then expressed in terms of the underlying total variation.

5.3 Decomposition with respect to a reference measure

The theorem is usually stated relative to a chosen reference measure. Different choices of reference measure can lead to different decompositions. In applications, the reference is often the most natural base measure, such as Lebesgue measure in real analysis or a probability measure in stochastic contexts.

5.4 Extensions to vector measures

For vector measures, decomposition theory becomes more delicate because values lie in a Banach space or similar structure. Under suitable hypotheses, one can still separate absolutely continuous and singular behavior. These results are useful in advanced integration theory and the study of operator-valued measures.

6 Examples

Concrete examples show how the theorem separates familiar measures into distinct parts. On the real line, one often sees mixtures of density, discrete atoms, and singular continuous behavior. Such examples make the abstract statement more intuitive.

6.1 Decomposition of a measure with density and atoms

A measure on the real line may consist of a smooth density plus a finite sum of point masses. Relative to Lebesgue measure, the density term is absolutely continuous, while the point masses are singular. The theorem identifies these pieces automatically.

6.2 Purely atomic measures

A purely atomic measure concentrates all its mass on countably many points. Relative to Lebesgue measure, it is singular because point sets are null for Lebesgue measure. This example illustrates the simplest kind of singular measure.

6.3 Measures singular to Lebesgue measure

Some measures are supported on sets of Lebesgue measure zero but are not atomic. Classic examples include measures carried by fractal-like subsets of the line. These show that singularity is broader than discreteness.

6.4 Mixed measures on the real line

A mixed measure may combine an absolutely continuous density, atomic contributions, and a singular continuous component. The Lebesgue decomposition theorem isolates the part with a density from the rest, and further refinements may separate atomic and non-atomic singular behavior. This layered structure is common in real-variable analysis.

7 Applications

The theorem has broad use across analysis and probability. It clarifies how one measure can be encoded by a density while also accounting for exceptional components. This makes it a foundational tool in areas where measures represent distributions or physical quantities.

7.1 Probability theory

In probability, the theorem helps classify random variables and distributions into continuous and discrete parts. It is used to describe mixtures of distributions and to identify when a probability law admits a density with respect to a reference measure. This is especially useful in studying transformations of random variables.

7.2 Functional analysis

Measure decomposition appears in duality theory, representation theorems, and the study of linear functionals on spaces of functions. The theorem helps separate regular behavior from singular contributions. It is often part of the machinery behind integral representations.

7.3 Harmonic analysis

In harmonic analysis, measures are often decomposed to analyze Fourier transforms and related operators. The distinction between absolutely continuous and singular components can affect convergence, regularity, and spectral properties. The theorem provides a clean framework for such distinctions.

7.4 Statistics and density estimation

In statistics, the theorem underlies the concept of a probability density with respect to a baseline measure. It also explains why some data-generating mechanisms produce a mixture of continuous and discrete features. Density estimation methods typically focus on the absolutely continuous portion of an empirical or model-based distribution.

Several classical results in measure theory are closely tied to the Lebesgue decomposition theorem. These include decompositions for signed measures and structural results about sets supporting different parts of a measure. Together they form a coherent toolkit for measure comparison.

8.1 Jordan decomposition

The Jordan decomposition expresses a signed measure as the difference of two nonnegative measures. It separates positive and negative variation and is often used before applying more refined decomposition results. It is distinct from, but compatible with, Lebesgue decomposition.

8.2 Hahn decomposition

The Hahn decomposition theorem partitions a space into positive and negative sets for a signed measure. It provides the basis for the Jordan decomposition and helps organize signed measure theory. Its role is foundational in handling measures with sign.

8.3 Singular and absolutely continuous parts of measures

These are the two components produced by the Lebesgue decomposition theorem. The absolutely continuous part is governed by a density relative to the reference measure, while the singular part is concentrated on a null set. Their separation is the theorem’s central achievement.

8.4 Direct sum decompositions in analysis

The theorem exemplifies a broader theme in analysis: a space or object can often be split into complementary parts. Such direct sum ideas appear in linear algebra, functional analysis, and operator theory. The measure-theoretic decomposition provides a canonical example of this structural viewpoint.