1 Definition and basic concepts

A nowhere dense set is a subset of a topological space that is small in the sense that it fails to occupy any open region in a robust way. The notion is central in topology and analysis because it captures “thinness” using only open sets, closure, and interior, without relying on distance or measure.

1.1 Topological setting

The concept is defined in any topological space. This generality is useful because it applies not only to subsets of the real line or Euclidean space, but also to abstract spaces arising in analysis, geometry, and functional analysis. The topology determines what counts as open, and therefore what it means for a set to be large enough to contain an open piece.

1.2 Interior and closure

Two standard operations are used in the definition: interior and closure. The interior of a set consists of all points around which some open neighborhood lies entirely inside the set. The closure is the smallest closed set containing the original set, or equivalently the set together with all its limit points. These operations measure how a set sits inside the surrounding space.

1.3 Formal definition of nowhere dense

A subset is nowhere dense if the interior of its closure is empty. In symbols, a set \(A\) is nowhere dense when \(\operatorname{int}(\overline{A})=\varnothing\). This means that even after adding all limit points of the set, the resulting closed set still contains no nonempty open subset.

1.4 Equivalent formulations

Several equivalent descriptions are often used, depending on context and convenience.

1.4.1 Closure has empty interior

The most direct formulation is that the closure of the set has empty interior. This restates the definition and emphasizes that the set remains topologically thin even after taking limit points into account.

1.4.2 No nonempty open subset is contained in the closure

Another equivalent form is that no nonempty open set lies inside the closure of the set. If an open region were contained in the closure, then the closure would have nonempty interior, contradicting nowhere density.

1.4.3 Density of the complement in every open set

A set is nowhere dense precisely when, in every nonempty open set, one can find a smaller nonempty open set disjoint from the closure of the set. In this sense, the complement of the closure is dense in each open region, showing that the set does not dominate any local part of the space.

2 Fundamental properties

Nowhere dense sets behave well under standard set-theoretic and topological operations, and these properties explain why they are useful in category arguments.

2.1 Subsets of nowhere dense sets

Any subset of a nowhere dense set is itself nowhere dense. If a larger set has closure with empty interior, then a smaller set cannot have a closure that becomes topologically thicker than the original one.

2.2 Behavior under closure

A set is nowhere dense if and only if its closure is nowhere dense. Since the definition already depends on the closure, passing to the closure does not change the property. This makes closed nowhere dense sets especially important, as they represent the same notion in a more rigid form.

2.3 Finite unions

The union of finitely many nowhere dense sets is nowhere dense. This reflects the idea that a finite combination of thin sets remains thin. The property fails for arbitrary countable unions, which is one reason Baire category theory is needed.

2.4 Relationship to dense sets

A nowhere dense set may still be dense in a small part of a space, but it cannot contain any open region after closure is taken. Thus density and nowhere density are distinct notions: density concerns approximation everywhere, while nowhere density concerns the absence of local thickness.

2.5 Comparison with empty interior

Every nowhere dense set has empty interior, but the converse need not hold. A set can have empty interior while its closure still contains an open set. For example, certain dense sets in an interval have no interior but are not nowhere dense because their closure is the whole interval.

3 Examples

Examples show that nowhere dense sets can be very simple or highly structured, and they appear in both classical and abstract settings.

3.1 Finite and countable sets in Euclidean spaces

In Euclidean spaces, finite sets are nowhere dense. More generally, countable subsets of \(\mathbb{R}^n\) are nowhere dense if they are closed and have no accumulation large enough to fill any open region; in particular, many familiar countable sets have closures with empty interior. The key point is that isolated or sparse point sets do not contain intervals, balls, or other open pieces.

3.2 The Cantor set

The Cantor set is a standard example of a closed nowhere dense subset of \([0,1]\). It is uncountable, perfect, and highly structured, yet it contains no interval and its closure is itself. Its example shows that “small” in the topological sense does not mean finite or countable.

3.3 Closed sets with empty interior

Any closed set with empty interior is nowhere dense. This class includes many boundary-type sets, such as certain fractal sets and geometric surfaces in higher-dimensional spaces. Closedness makes the verification immediate, since the closure is the set itself.

3.4 Discrete subsets

Discrete subsets of spaces like \(\mathbb{R}^n\) are typically nowhere dense when they have no accumulation point inside an open region. Because each point is isolated, such sets do not contain any open neighborhood around their points. Their closures often coincide with the sets themselves or add only isolated limit points.

3.5 Examples in function spaces

In spaces of functions equipped with standard topologies, many naturally occurring constraint sets are nowhere dense. For instance, classes defined by rigid pointwise equalities or special algebraic restrictions can fail to contain any open neighborhood in the function-space topology. Such examples are important in genericity arguments, where one wants to show that certain exceptional behaviors are rare.

4 Non-examples

Not every set with a small-looking description is nowhere dense. The defining test is always whether the closure has empty interior.

4.1 Open sets

No nonempty open set is nowhere dense. Its closure contains the open set itself or at least an open region, so the interior of the closure is nonempty.

4.2 Dense open sets

Dense open sets are maximally opposite to nowhere dense sets in a local topological sense. Their closure is the whole ambient space, so they have large interior unless the entire space is trivial. They are used frequently in arguments showing that a property holds on a large set of points.

4.3 Sets with nonempty interior

Any set that already contains a nonempty open subset is not nowhere dense. The presence of an interior point with an open neighborhood immediately prevents the closure from being topologically thin.

4.4 Dense subsets of intervals and Euclidean spaces

Dense subsets such as the rational numbers in an interval are not nowhere dense, even though they have empty interior. Their closure is the entire interval, whose interior is nonempty. This illustrates why empty interior alone is insufficient for nowhere density.

5 Relation to Baire category

Nowhere dense sets play a central role in the distinction between category-small and category-large sets.

5.1 Meagre sets

A meagre set is a countable union of nowhere dense sets. Such sets are also called first category sets. The notion generalizes the idea of being small by allowing countably many thin pieces rather than just one.

5.2 Nowhere dense sets versus meagre sets

Every nowhere dense set is meagre, but not every meagre set is nowhere dense. The difference is that meagre sets may be built from many nowhere dense layers, while a nowhere dense set is thin all at once. This distinction is important in finer category arguments.

5.3 Countable unions of nowhere dense sets

Countable unions of nowhere dense sets can be surprisingly large in appearance, even though they remain category-small. The rationals in \(\mathbb{R}\) are the classic example: each singleton is nowhere dense, and their countable union is meagre. This shows that category-smallness is compatible with density.

5.4 Baire category theorem

The Baire category theorem states, in suitable spaces such as complete metric spaces, that the space cannot be written as a countable union of nowhere dense sets. This theorem is a major tool in analysis because it guarantees the existence of points where “generic” behavior must occur. It is often used to prove that certain exceptional sets are small in the category sense.

6 Applications

Nowhere dense sets are used to formalize exceptional behavior and to support genericity arguments across analysis and topology.

6.1 Real analysis

In real analysis, nowhere dense sets help distinguish typical behavior from exceptional behavior in the real line. They are useful in studying sets of discontinuities, oscillation phenomena, and subsets that fail to contain intervals. They also appear in constructions showing that certain properties are rare.

6.2 Topology

In topology, nowhere dense sets are a standard way to describe sets that do not locally fill space. They are often used in decompositions of spaces into large and small parts, and in proving that particular subspaces are negligible from a structural viewpoint.

6.3 Functional analysis

In functional analysis, nowhere dense sets are important in spaces of operators, functions, and distributions. They help identify constraint sets that are not stable under perturbation and therefore represent nongeneric behavior. Many existence proofs use Baire category to show that the complement of a nowhere dense or meagre set is large.

6.4 Generic properties in analysis

A property is often called generic if it holds on a residual set, meaning the complement is meagre. Since nowhere dense sets are the building blocks of meagre sets, they provide the basic notion of exceptional behavior in this framework. This is especially useful when proving that “most” objects in a function space satisfy a certain property.

Several nearby concepts help clarify the position of nowhere dense sets among other forms of topological smallness.

7.1 Somewhere dense sets

A set is somewhere dense if its closure has nonempty interior. This is the direct opposite of nowhere dense, at least in the sense that the closure does contain an open region. Many common sets are somewhere dense, especially dense subsets of spaces with nonempty interior.

7.2 Dense-in-itself sets

A dense-in-itself set has no isolated points. This property is different from nowhere density: a set may be dense-in-itself and still be nowhere dense, as in the Cantor set. The notion concerns local point structure rather than the size of the closure’s interior.

7.3 Thin sets in other contexts

Other branches of mathematics use “thin” or “small” in analogous ways, sometimes through measure, dimension, algebraic complexity, or combinatorial sparsity. Although these notions are not identical to nowhere density, they often serve a similar purpose: identifying sets that are exceptional relative to a given structure.

7.4 Nowhere dense relative to a subspace

A set may be nowhere dense in a subspace even if it is not nowhere dense in the larger ambient space, and vice versa. Relative nowhere density is determined by the topology inherited from the subspace. This local viewpoint is often useful when working inside manifolds, subspaces of function spaces, or subsets with induced topology.

8 References and further reading

Foundational treatments of nowhere dense sets appear in general topology, real analysis, and the theory of Baire spaces. Standard references present the topic as part of the broader study of category and generic properties.

8.1 Classical topology texts

Classical topology books typically introduce nowhere dense sets alongside interior, closure, and boundary. They place the concept within the study of category, connectedness, and completeness, often using it to motivate the Baire category theorem.

8.2 Measure and category references

Texts on measure and category compare nowhere dense sets with null sets and other notions of smallness. These references are especially helpful for understanding the difference between topological smallness and measure-theoretic smallness, as well as their points of overlap and divergence.