Metrization Theorems

Metrization theorems are the bridges that connect the abstract landscape of general topology to the more intuitive world of metric spaces. These results tell you precisely when the open sets of a topological space can be described using a distance function, transforming qualitative notions of "closeness" into quantitative measurements. Understanding these theorems is crucial because it allows you to leverage the powerful analytical tools of metric spaces—like sequences, completeness, and uniform continuity—within a much broader topological context.

Prerequisites for Metrizability

Before tackling the main theorems, we must establish the necessary topological "symptoms" that suggest a space might be metrizable—meaning there exists a metric $d$ on the set $X$ whose open balls generate the given topology. Three key properties stand out.

First, a metrizable space must be Hausdorff ($T_2$). Distinct points can be separated by disjoint open sets because, in a metric space, you can place open balls of sufficiently small radius around each point. More importantly, any metrizable space is completely normal and Tychonoff ($T_{3\frac{1}{2}}$). This includes being regular ($T_3$): points and disjoint closed sets can be separated by open sets. The metric itself provides the mechanism for this separation.

Second, metrizable spaces have a strong countability property related to bases. A basis for a topology is a collection of open sets such that every open set is a union of basis elements. A second-countable space has a countable basis. While not all metric spaces are second-countable (consider an uncountable set with the discrete metric), a separable metric space—one with a countable dense subset—is second-countable. This link is a key insight for Urysohn's theorem.

Finally, we consider a more refined notion than a basis: a local base at a point $x$ is a collection of open neighborhoods of $x$ such that every neighborhood of $x$ contains one from the collection. A space is first-countable if every point has a countable local base. All metric spaces are first-countable because the balls $B(x,1/n)$ for $n\in \mathbb{N}$ form a countable local base. First-countability is a necessary condition for metrizability, but far from sufficient.

Urysohn's Metrization Theorem

Urysohn's metrization theorem provides a classic and relatively accessible sufficient condition for metrizability. It states: Every second-countable, regular topological space is metrizable.

The power of this theorem lies in its combination of two purely topological conditions to guarantee the existence of a compatible metric. The proof is constructive and illuminating. It leverages second-countability to obtain a countable basis $\{B_n\}$, and then uses regularity to apply Urysohn's Lemma. This lemma, a cornerstone of topology, guarantees that for any closed set $C$ and an open set $U$ containing it, there exists a continuous function $f:X\to [0,1]$ with $f(C)=0$ and $f(X\setminus U)=1$.

The construction proceeds by selecting, for each basis element $B_n$ and a point $x\in B_n$, a continuous function that is $0$ at $x$ and $1$ outside $B_n$. The countability of the basis allows us to assemble a countable family of such functions $\{f_n\}$. These are then used to define an embedding $F:X\to [0,1{]}^{\omega }$ into the Hilbert cube (the countable product of unit intervals) via $F(x)=(f_1(x),f_2(x),\dots )$. The Hilbert cube is a known metrizable space. The theorem shows $F$ is a homeomorphism onto its image, thereby transferring the metric from the Hilbert cube back to $X$. This process shows that any such space is not only metrizable but can be embedded in a familiar, concrete metric space.

The Nagata-Smirnov Metrization Theorem

While Urysohn's theorem is elegant, second-countability is a restrictive global condition. The Nagata-Smirnov metrization theorem provides a complete characterization of metrizability, extending to non-separable spaces. Its statement requires a new concept: a $\sigma$-locally finite base.

A collection of sets is locally finite if every point has a neighborhood intersecting only finitely many sets in the collection. A collection is $\sigma$-locally finite if it is a countable union of locally finite families. The theorem then states: *A topological space is metrizable if and only if it is regular and has a $\sigma$-locally finite basis.*

This theorem sharpens our understanding immensely. The "if" direction (sufficiency) is the hard part of the proof, generalizing the embedding idea of Urysohn but with intricate technical care to handle the lack of global countability. The "only if" direction (necessity) is also insightful: it shows that any metric space $(X,d)$ possesses such a basis. One can construct it by considering, for each $n\in \mathbb{N}$, the collection of all balls of radius $1/n$. While this is uncountable, using the fact that $X$ can be covered by a locally finite family of sets of diameter less than $1/n$ (a property following from paracompactness of metric spaces), one can refine these covers to build the required $\sigma$-locally finite basis.

This theorem is often presented alongside the Bing metrization theorem, which uses a $\sigma$-discrete basis (where each family is a collection of pairwise disjoint open sets). Together, they form the definitive classical answer to the metrization problem.

Common Pitfalls

1. Assuming regularity is automatic. A common error is to believe "Hausdorff" or "normal" is enough. Urysohn's theorem explicitly requires regularity ($T_3$). There exist Hausdorff, second-countable spaces that are not regular (like certain quotient spaces of $\mathbb{R}$), and these are not metrizable. Always verify both regularity and the countability condition.
2. Confusing necessary and sufficient conditions. First-countability is necessary for metrizability, but not sufficient. The space $\mathbb{R}$ with the lower-limit topology (Sorgenfrey line) is first-countable and regular but not metrizable (as it is separable but not second-countable). Do not mistake a prerequisite for a guarantee.
3. Over-applying Urysohn's theorem. Urysohn's theorem states that second-countable + regular $⟹$ metrizable. It does not state that metrizable $⟹$ second-countable. Many metric spaces (like uncountable discrete spaces) are metrizable but not second-countable. The correct converse for separable metric spaces is: metrizable + separable $⟹$ second-countable.
4. Misunderstanding the basis in Nagata-Smirnov. The condition is not simply having a basis that is a countable union of sets. It is a countable union of collections that are each locally finite. This is a subtle but crucial distinction regarding how the basis elements are distributed across the space, ensuring they don't "pile up" too much at any single point.

Summary

• Metrization theorems establish the conditions under which an abstract topological space admits a compatible metric, allowing the use of analytic tools from metric space theory.
• Urysohn's metrization theorem provides a relatively simple-to-check sufficient condition: a second-countable, regular ($T_3$) space is always metrizable. The proof constructively embeds the space into the Hilbert cube.
• The Nagata-Smirnov theorem gives a complete characterization: a space is metrizable if and only if it is regular and has a $\sigma$-locally finite basis. This is the definitive solution for general topological spaces.
• Key necessary conditions for metrizability include being Hausdorff, regular, normal, Tychonoff, and first-countable, but none of these alone or in certain combinations is sufficient.
• Always carefully distinguish between properties that are necessary for metrizability (like first-countability) and those that, in combination, are sufficient (like second-countable + regular).