1 Discrete Metric Definition

1.1 Metric on an Arbitrary Set

Let \(X\) be a nonempty set. The discrete metric is a function \(d:X\times X\to \mathbb{R}_{\ge 0}\) defined so that points are either identical (distance \(0\)) or distinct (distance \(1\)). Concretely, the standard discrete metric is \[ d(x,y)= \begin{cases} 0, & x=y,\\ 1, & x\ne y. \end{cases} \] This construction works for any underlying set \(X\), whether it is finite or infinite, countable or uncountable.

1.2 Distance Rule and Properties

To be a metric, \(d\) must satisfy three axioms: identity of indiscernibles, symmetry, and the triangle inequality. The discrete metric is designed to satisfy these requirements with minimal effort because it uses only the two values \(0\) and \(1\).

1.2.1 Identity of Indiscernibles

The identity of indiscernibles states that \(d(x,y)=0\) if and only if \(x=y\). For the discrete metric, \(d(x,y)=0\) occurs exactly when \(x=y\) by definition, and when \(x\ne y\) the distance is \(1\), never \(0\).

1.2.2 Symmetry

Symmetry requires \(d(x,y)=d(y,x)\) for all \(x,y\in X\). Since the rule depends only on whether \(x\) equals \(y\), swapping the arguments does not change the outcome: both \(d(x,y)\) and \(d(y,x)\) are \(0\) when \(x=y\), and \(1\) otherwise.

1.2.3 Triangle Inequality

The triangle inequality requires \(d(x,z)\le d(x,y)+d(y,z)\) for all \(x,y,z\in X\). In the discrete metric, the left-hand side is either \(0\) or \(1\). If \(x=z\), then \(d(x,z)=0\) and the inequality holds trivially. If \(x\ne z\), then \(d(x,z)=1\), and the sum \(d(x,y)+d(y,z)\) is always at least \(1\): it cannot happen that both terms are \(0\) (that would force \(x=y\) and \(y=z\), hence \(x=z\)). Therefore the triangle inequality is satisfied.

1.3 Special Cases and Examples

1.3.1 One-Point Set

If \(X=\{p\}\) has one element, the discrete metric is uniquely determined because there is only one pair \((p,p)\). The distance is \(d(p,p)=0\), producing a metric space with no nontrivial distances.

1.3.2 Finite Sets

For a finite set \(X\) with \(n\) elements, the discrete metric assigns distance \(1\) to every distinct pair. As a result, the metric space has a simple geometry: any two different points are equally separated, with no “intermediate” distances.

1.3.3 Infinite Sets

For an infinite set \(X\), the same distance rule applies. Even though the space may be infinite and may contain infinitely many points, the metric still forces uniform separation between any two distinct elements, which strongly simplifies questions about convergence, Cauchy sequences, and topology.

2 Discrete Metric Spaces

2.1 Open Balls in the Discrete Metric

Let \(B(x,r)=\{y\in X: d(x,y)<r\}\) be the open ball of radius \(r\) centered at \(x\). Because \(d\) takes only the values \(0\) and \(1\), the ball structure is determined entirely by whether \(r\) is less than or at least \(1\).

2.1.1 Balls of Radius Less Than 1

If \(0<r<1\), then \(d(x,y)<r\) can only hold when \(d(x,y)=0\), since the only nonzero distance is \(1\). Thus, \[ B(x,r)=\{x\} \quad\text{for }0<r<1. \] So small open balls contain exactly one point.

2.1.2 Balls of Radius Greater Than or Equal to 1

If \(r\ge 1\), then \(d(x,y)<r\) holds both when \(y=x\) (distance \(0\)) and when \(y\ne x\) (distance \(1\)), because \(1<r\) when \(r>1\) and when \(r=1\) the inequality \(d(x,y)<1\) excludes the case \(d=1\). Using the strict inequality in the definition of open ball:

  • For \(r>1\), every point is included, so \(B(x,r)=X\).
  • For \(r=1\), the condition \(d(x,y)<1\) again forces \(y=x\), so \(B(x,1)=\{x\}\).

Equivalently, with the strict inequality convention, balls are either singleton sets (for \(r\le 1\)) or the whole space (for \(r>1\)).

2.2 Open Sets and Topology

The open sets induced by the metric are those that can be written as unions of open balls. In the discrete metric, the behavior of open balls makes the topology exceptionally simple.

2.2.1 Every Subset Is Open

Because for any \(x\in X\) and any \(r\) with \(0<r<1\), the ball \(B(x,r)\) equals \(\{x\}\), each singleton \(\{x\}\) is open. An arbitrary subset \(U\subseteq X\) can then be expressed as \[ U=\bigcup_{x\in U}\{x\}, \] a union of open sets, hence itself open. Therefore every subset of \(X\) is open.

2.2.2 Discrete Topology Equivalence

The topology in which every subset is open is called the discrete topology. The discrete metric induces exactly this topology. Conversely, given a topology that declares all subsets open, there exists a compatible metric structure; for the metric space viewpoint, the discrete metric is the canonical example.

2.3 Closed Sets and Complements

In any metric space, a set is closed if its complement is open. Since openness is universal in a discrete metric space, the closed sets can be described just as directly.

2.3.1 Characterization of Closed Sets

A subset \(F\subseteq X\) is closed precisely when \(X\setminus F\) is open. But every subset is open, so every complement is open, and therefore every subset is closed as well.

2.3.2 Boundary Behavior

The boundary \(\partial A\) of a set \(A\) consists of points where every neighborhood intersects both \(A\) and its complement. In a discrete metric space, neighborhoods of a point \(x\) can be chosen as the singleton \(\{x\}\) (using a radius \(r\) with \(0<r<1\)). If \(x\in A\), then \(\{x\}\) lies entirely in \(A\), so it cannot meet the complement; similarly if \(x\notin A\). Consequently, the boundary of \(A\) is empty unless the set is neither open nor closed, which never happens here. Thus \(\partial A=\varnothing\) for all \(A\subseteq X\).

3 Convergence and Continuity

3.1 Convergent Sequences

Convergence is defined by the behavior of distances to a candidate limit point. In the discrete metric, that behavior becomes all-or-nothing.

3.1.1 When a Sequence Converges

A sequence \((x_n)\) converges to \(x\) if for every \(\varepsilon>0\), there exists \(N\) such that for all \(n\ge N\), \(d(x_n,x)<\varepsilon\). Choose \(\varepsilon\) with \(0<\varepsilon\le 1\). Then \(d(x_n,x)<\varepsilon\) implies \(d(x_n,x)=0\), hence \(x_n=x\). So convergence to \(x\) occurs exactly when the sequence is eventually constant at \(x\).

3.1.2 Eventually Constant Sequences

Summarizing the previous point: \[ x_n\to x \quad\Longleftrightarrow\quad \exists N\ \forall n\ge N,\ x_n=x. \] In other words, the only convergent sequences are those that “settle” on a single point and never leave it afterward.

3.2 Cauchy Sequences

Cauchy sequences generalize convergence by requiring that points in the tail of the sequence become arbitrarily close to one another.

3.2.1 Cauchy Criterion in Discrete Metrics

A sequence \((x_n)\) is Cauchy if for every \(\varepsilon>0\), there exists \(N\) such that for all \(m,n\ge N\), \(d(x_m,x_n)<\varepsilon\). Again take \(0<\varepsilon\le 1\). The inequality forces \(d(x_m,x_n)=0\), so \(x_m=x_n\) for all \(m,n\ge N\). Therefore, a sequence is Cauchy if and only if it is eventually constant.

3.2.2 Relation to Completeness

A metric space is complete if every Cauchy sequence converges. Since in the discrete metric a Cauchy sequence is eventually constant, it automatically converges to its eventual value. Hence discrete metric spaces are complete.

3.3 Continuity of Maps

3.3.1 Automatically Continuous Functions

Let \(f:(X,d)\to (Y,\rho)\) be a function from a discrete metric space into any metric space \((Y,\rho)\). Continuity at a point \(x\) requires that whenever \(x_n\to x\), the images satisfy \(f(x_n)\to f(x)\). But \(x_n\to x\) implies \(x_n\) equals \(x\) for large \(n\), so \(f(x_n)\) equals \(f(x)\) for large \(n\) and thus converges. Consequently, every function out of a discrete metric space is continuous.

3.3.2 Homeomorphisms and Isomorphisms

If both domain and codomain carry discrete metrics, then topologically they have the same property (all sets are open). A bijection \(f:X\to Y\) is a homeomorphism exactly when it is continuous with continuous inverse; in the discrete setting this reduces to the basic fact that every bijection is continuous and every bijection has a continuous inverse. In addition, an isometry is a bijection preserving the discrete distance; because the only nonzero distance is \(1\), any bijection between discrete metric spaces is automatically an isometry.

4 Metric Space Structure and Separability

4.1 Completeness of Discrete Metric Spaces

As established via Cauchy sequences, discreteness forces Cauchy sequences to become constant in the tail, which ensures convergence. Therefore every discrete metric space is complete. This completeness holds regardless of whether the underlying set is finite or infinite.

4.2 Compactness Characteristics

Compactness in metric spaces can be characterized through sequences or open covers. Discrete metric spaces have a stark compactness pattern.

4.2.1 Finite Discrete Spaces

If \(X\) is finite, then every open cover has a finite subcover automatically, since there are only finitely many points. Hence a finite discrete metric space is compact.

4.2.2 Infinite Discrete Spaces

If \(X\) is infinite, compactness fails. One way to see this is through sequences: in an infinite discrete space, one can choose a sequence of distinct points \((x_n)\). No such sequence converges (convergence would require eventual constancy), and in metric spaces, the existence of nonconvergent sequences is compatible with noncompactness. Another direct argument uses open covers: the collection of singleton sets \(\{\{x\}:x\in X\}\) is an open cover, but no finite subcollection can cover an infinite set. Thus infinite discrete metric spaces are not compact.

4.3 Total Boundedness

Total boundedness means that for every radius, the space can be covered by finitely many balls of that radius.

4.3.1 Finite vs Infinite Cases

Fix \(r\) with \(0<r\le 1\). In that regime, open balls are singletons. Covering \(X\) by finitely many such balls would require \(X\) to be finite. Hence a discrete metric space is totally bounded if and only if it is finite. For \(r>1\), a single ball already covers the whole space, but total boundedness requires the property for all radii, including those \(\le 1\).

4.4 Connectedness

Connectedness asks whether the space can be split into two disjoint nonempty separated open sets. Discrete spaces resist such splittings in the opposite direction: they split everywhere.

4.4.1 Strong Disconnectedness Intuition

Every singleton is open and closed, so for any two distinct points \(x\ne y\), the sets \(\{x\}\) and \(X\setminus\{x\}\) form a separation: both are open, disjoint, and nonempty. Therefore the space is totally disconnected: no nontrivial connected subsets exist. In practical terms, discrete metric spaces behave as if points “do not touch” one another at all.

5 Transformations and Variants

5.1 Scaling Metrics (d’ = c·d)

If \(d\) is the discrete metric and \(c&gt;0\), then \(d&#039;(x,y)=c\,d(x,y)\) defines a new metric. Here \(d&#039;(x,x)=0\) and \(d&#039;(x,y)=c\) when \(x\ne y\). Scaling preserves many metric properties that depend only on the structure of convergence and topology; in particular, the induced topology remains discrete because the set of distances from a point to other points is still separated into “zero” and “positive constant.”

5.2 Truncated or Modified Discrete Metrics

One can alter the discrete metric while keeping the basic idea: use a fixed positive distance for distinct points and perhaps compress it through a monotone transformation. For instance, applying a strictly increasing function \(\phi\) with \(\phi(0)=0\) to \(d\) yields \(\phi(d(x,y))\), which remains a metric provided \(\phi(1)>0\). Such modifications generally preserve the discrete topology and the same convergence behavior (eventual constancy), since the only way to get arbitrarily small distances is to reach the same point.

5.3 Induced Pseudometrics Versus Metrics

A pseudometric allows \(d(x,y)=0\) for distinct points. If one changes the discrete rule so that some distinct pairs have distance \(0\), the resulting function may be a pseudometric rather than a metric. The topological and convergence behavior can then change: points at zero distance are indistinguishable for the purposes of neighborhoods, even though they may be distinct elements. The standard discrete metric avoids this by making \(d(x,y)=0\) exactly when \(x=y\), keeping it a genuine metric.

In more elaborate constructions, such as forming products of sets or considering related spaces, one can define product-like distances by combining component distances. When component metrics are discrete, the resulting metric often reflects whether at least one coordinate differs: distances become sums, maxima, or weighted combinations of the component distances. Conceptually, the discrete nature tends to simplify the distance comparison to whether coordinates match or differ, though the exact formula determines the specific value assigned and whether the metric is exactly discrete or a close variant.