1 Definition and basic ideas

A null set is a set that is negligible with respect to a chosen measure. Informally, it is “small” not by cardinality, but by size as measured in the sense of measure theory. In the real line, null sets are typically the sets of Lebesgue measure zero. More generally, the notion depends on the underlying measure space.

Null sets are important because many results in analysis are not required to hold at every point. Instead, they may fail only on a null set, which is regarded as insignificant for most purposes. This leads to the widely used phrase “almost everywhere.”

1.1 Measure zero

In the simplest setting, a set \(E \subseteq \mathbb{R}\) has measure zero if for every \(\varepsilon > 0\), it can be covered by countably many intervals whose total length is less than \(\varepsilon\). Such a set may contain infinitely many points, even uncountably many, while still occupying no measurable length.

This notion captures the idea that some sets are too thin or sparse to contribute to area, length, or volume in integration. A set of measure zero has no effect on the value of an integral in the Lebesgue sense when the integrand is altered only on that set.

1.2 Null sets in measure spaces

In a measure space \((X,\Sigma,\mu)\), a null set is any set \(N \in \Sigma\) with \(\mu(N)=0\). When the measure is understood, the term usually refers to sets with zero measure under that particular measure.

Different measures can assign different sizes to the same set. A set that is null for one measure may be large for another. For example, a set of Lebesgue measure zero on \(\mathbb{R}\) need not be small with respect to counting measure.

1.3 Equivalent formulations

For subsets of the real line, several formulations are commonly used. A set has measure zero if it can be covered by intervals with arbitrarily small total length. In complete measure spaces, any subset of a null measurable set is also null. In Euclidean spaces, analogous definitions use rectangles, cubes, or more general coverings adapted to the relevant measure.

These equivalent viewpoints emphasize that nullness is a structural property rather than a geometric one. The precise formulation may vary, but the core idea remains the absence of measurable size.

2 Examples of null sets

Null sets appear in many familiar forms. Some are very simple, while others are highly intricate and fractal-like.

2.1 Finite and countable sets

Any finite set has measure zero in the usual Lebesgue measure on \(\mathbb{R}\). More generally, any countable set is null, since each point can be covered by an interval of very small length and the resulting lengths can be summed to an arbitrarily small total.

This includes familiar examples such as the integers, rational numbers, and algebraic numbers. Although these sets may be dense or infinite, they remain negligible from the viewpoint of length or area.

2.2 The empty set

The empty set is the simplest null set. Its measure is zero in every standard measure space. It is often used as the trivial example of a set of no size at all.

2.3 Cantor set

The Cantor set is a classical uncountable set of measure zero in \(\mathbb{R}\). It is formed by repeatedly removing the middle third of intervals, leaving a compact, perfect, nowhere dense set. Despite containing uncountably many points, it has zero Lebesgue measure.

The Cantor set is especially important because it shows that null sets can be complicated and not merely finite or countable. It is often used to illustrate the distinction between size in the sense of measure and size in the sense of cardinality.

2.4 Other classical examples

Many graphs of well-behaved functions can produce null sets when viewed in certain dimensions. For example, the graph of a continuous function in the plane has area zero, though it may extend infinitely. Certain exceptional sets arising in analysis, such as sets where a function fails to be differentiable, are often null.

In higher dimensions, lower-dimensional geometric objects like curves, surfaces, and hyperplanes are typically null with respect to the ambient Lebesgue measure.

3 Properties of null sets

Null sets have robust closure properties, which make them easy to handle in proofs and applications. They form an ideal-like family under common operations.

3.1 Subsets of null sets

Any subset of a null set is also null, provided it is measurable or the measure space is complete enough to assign measure zero to all such subsets. This property reflects the idea that taking away points cannot create measurable size.

In complete measure spaces, this inheritance is automatic. It is one reason null sets are useful for describing exceptions: once a set is negligible, all of its parts are negligible as well.

3.2 Countable unions of null sets

A countable union of null sets is null. This follows from countable subadditivity of measure. If each set has measure zero, then the total measure of their union remains zero.

This property is central in analysis because many exceptional sets arise as countable unions of simpler exceptional sets. It allows one to combine many small failures into a single negligible set.

3.3 Translation and scaling

In Euclidean spaces with Lebesgue measure, nullness is preserved under translations and nonzero scalar multiplication. If a set has measure zero, then any shifted or rescaled copy also has measure zero. More generally, rigid motions and certain smooth transformations preserve null sets under appropriate conditions.

This invariance shows that nullness is geometric rather than dependent on location. A set is small everywhere, not just at one particular position or scale.

3.4 Complements and non-null sets

The complement of a null set need not be large in every intuitive sense, but it has full measure within the ambient space when the ambient space itself has finite measure. In such settings, null sets represent the entire loss from the whole space.

A non-null set is one with positive measure. Such sets are, in a precise sense, large enough to contribute to integrals, probabilities, or geometric size.

4 Null sets in real analysis

Null sets are woven into the language of real analysis, where they help formalize the idea that some irregularities are too rare to matter.

4.1 Almost everywhere statements

A property holds almost everywhere if the set of points where it fails is a null set. This is stronger than saying the property holds at many points, but weaker than holding everywhere.

Almost everywhere statements are common because many analytic results are naturally insensitive to changes on negligible sets. Functions that differ only on a null set are often treated as essentially the same for integration and many limit processes.

4.2 Exceptional sets in integration

In integration theory, null sets are the places where functions may behave badly without affecting the integral. A function altered on a null set has the same Lebesgue integral as the original, assuming measurability and integrability are preserved.

This feature distinguishes Lebesgue integration from more elementary forms of integration. It allows a more flexible treatment of discontinuities, spikes, and other irregular behavior.

4.3 Relation to continuity and differentiability

Many real-valued functions are continuous or differentiable except on null sets. For example, monotone functions on intervals are differentiable almost everywhere. Similarly, functions of bounded variation have derivatives almost everywhere in a suitable sense.

These results show that large classes of functions may be highly irregular at some points while still being well behaved in the aggregate. Null sets identify the exceptional locations without undermining the main theorem.

5 Null sets in measure theory

Measure theory provides the formal framework in which null sets are defined and studied. The concept interacts closely with measurability, completion, and outer measure.

5.1 Measurable null sets

A measurable null set is a measurable set with measure zero. In most standard settings, this is the primary type of null set considered. Such sets are especially convenient because measure-theoretic operations are well defined on them.

Not every subset of a null measurable set must be measurable unless the space is complete. This distinction is important in advanced measure theory.

5.2 Completeness of a measure space

A measure space is complete if every subset of a null set is measurable and null. Completeness ensures that no negligible set is missing from the measurable structure.

The completion of a measure space adds all subsets of null sets to the measurable collection. This process is often used to eliminate technical gaps and make the theory more robust.

5.3 Outer measure and completion

Outer measure extends the idea of size to all subsets, measurable or not. A set of outer measure zero is null in a broad sense, even before measurability is imposed. This notion is closely related to the construction of Lebesgue measure.

Completion then uses null sets to enlarge the measurable sigma-algebra. In this way, null sets help bridge the gap between abstract measure spaces and fully usable analytic frameworks.

6 Null sets in probability

In probability theory, null sets correspond to events of probability zero. They are the probabilistic analogue of measure-zero sets.

6.1 Events of probability zero

An event with probability zero is called a null event. Such events may still be possible, depending on the model. For example, when choosing a point uniformly from an interval, the event of selecting any particular point has probability zero.

Probability zero does not necessarily mean impossibility. It means that the event is negligible in the probability measure.

6.2 Almost sure statements

A statement is true almost surely if it holds except on a probability-zero set of outcomes. This is one of the most common modes of assertion in probability theory and stochastic processes.

Almost sure results are the probabilistic counterpart of almost everywhere statements in analysis. They are strong enough for most applications while allowing exceptional sample paths or outcomes.

6.3 Zero-probability versus impossibility

An event of probability zero may still occur in continuous probability models. This distinction is essential: probability measures describe long-run frequency or likelihood, not metaphysical impossibility.

For example, a randomly chosen real number from an interval almost surely is irrational, even though rational numbers remain possible. The set of rational outcomes is null, not nonexistent.

7 Applications

Null sets are indispensable across analysis and its applications. They simplify statements, clarify exceptional behavior, and support flexible equivalence relations among functions.

7.1 Lebesgue integration

Lebesgue integration is built to ignore changes on null sets. If two functions are equal almost everywhere, then they define the same integral whenever the integral exists. This makes the theory particularly well suited to piecewise-defined or irregular functions.

Null sets also help in convergence theorems, where pointwise behavior may fail on negligible exceptional sets. This flexibility is one of the major strengths of the Lebesgue framework.

7.2 Theorems holding almost everywhere

Many fundamental theorems in analysis conclude that a property holds almost everywhere rather than everywhere. This includes results about differentiation of integrals, behavior of monotone functions, and convergence of certain sequences of functions.

Such theorems are powerful because they describe the typical behavior of a function or process while acknowledging rare exceptions. Null sets provide the formal language for those exceptions.

7.3 Functional analysis and \(L^p\) spaces

In functional analysis, functions are often identified if they differ only on a null set. This leads to the construction of \(L^p\) spaces, where functions are treated as equivalence classes under almost everywhere equality.

This viewpoint is essential because norms and inner products in these spaces depend on integrals, which ignore null sets. As a result, null sets are built into the foundational structure of many spaces of functions.

Several related notions are used alongside null sets, especially when comparing measure theory with topology and abstract analysis.

8.1 Measure zero sets

A measure zero set is a set whose measure is zero under the specified measure. In many contexts, this is simply another name for a null set. The term emphasizes the numerical value of the measure rather than the qualitative idea of negligibility.

8.2 Negligible sets

A negligible set is one that can be disregarded for the purposes of the theory at hand. In measure theory, this usually means a null set. The word highlights the practical role of these sets in simplifying arguments and definitions.

8.3 Meagre sets and category

Meagre sets are small in the sense of Baire category, not measure. They are countable unions of nowhere dense sets. A set may be meagre without being null, null without being meagre, both, or neither.

This distinction shows that “smallness” has multiple meanings in mathematics. Measure and category are related but fundamentally different ways of describing exceptional sets.

8.4 Almost everywhere equivalence

Two functions are almost everywhere equivalent if they agree outside a null set. This relation is central in measure theory, where many function spaces are defined using such equivalence classes.

Almost everywhere equivalence allows analysts to focus on essential behavior rather than pointwise anomalies. It is one of the main reasons null sets play such a prominent role in modern analysis.