Moore Plane

In the vast landscape of mathematical objects, few are as deceptively simple and profoundly instructive as the Moore plane. At first glance, it is merely the upper half of the Cartesian plane, a familiar setting. However, its true nature is revealed by a subtle twist in the definition of "nearness" on its boundary, a change that challenges our fundamental intuitions about space. This article delves into this classic counterexample to explore the gap between seemingly "well-behaved" spaces and the stricter conditions required for properties like normality and metrizability. We will first uncover the unique rules governing its structure in "Principles and Mechanisms," demonstrating how it can be both regular and Hausdorff yet fail the crucial test of normality. Following this, the "Applications and Interdisciplinary Connections" section will illuminate its role as a powerful tool for testing the limits of theorems in topology, analysis, and beyond, revealing why this strange world is so essential for a deeper understanding of mathematics.

So, we have been introduced to this curious mathematical object, the Moore plane. At first glance, it seems simple enough—it's just the upper half of the familiar Cartesian plane, including the $x$-axis. But as with so many things in mathematics, the real character of a space is revealed not by its points, but by how we define "nearness." The rules of what constitutes an "open set" or a "neighborhood" define the topology, the very fabric of the space. And for the Moore plane, these rules are where the beautiful strangeness begins.

Imagine the Moore plane, which we'll call $\Gamma$, as a flat landscape bordered by a long, straight coastline (the $x$-axis, which we'll call $L$). The landscape itself is the open upper half-plane, $U$.

For any point $p$ far from the coast, out in the "inland" region $U$, the rules of nearness are exactly what you'd expect. A neighborhood is just a standard open disk, a circular patch of land around $p$. As long as you don't try to draw a circle so big that it hits the coastline, everything feels normal. This part of the space is, for all intents and purposes, just a piece of the good old Euclidean plane.

The real fun begins when we approach the coastline $L$. What is a neighborhood of a point $q$ sitting right on the $x$-axis? Here, the rules change dramatically. You can't just draw a disk around $q$, because half of that disk would be "underwater" (in the lower half-plane, which isn't part of our space). The definition is more peculiar and far more interesting. A basic neighborhood of a point $q$ on the line is the point $q$ itself, plus an open disk from the upper half-plane $U$ that comes down and kisses the coastline precisely at $q$, like a perfectly formed bubble tangent to the line at that single point.

Think about what this means. Any open set that contains a point $q$ on the boundary must, by its very definition, include one of these "tangent bubbles" and therefore must contain points from the upper half-plane $U$. You cannot take a small step along the coastline from one point to another and remain within a single small bubble-neighborhood. Each point on the line has its own private set of bubbles reaching up into the plane, touching the line only at that point. In the topology of the Moore plane, the points on the line $L$ are profoundly isolated from each other. If you were an inhabitant of this world, to get from one coastal town to the next, you couldn't just walk along the beach; you would have to make a little excursion inland and come back. This tells us that the subspace topology on the line $L$ is the ​​discrete topology​​—every single point forms its own open set relative to the line itself.

This dual-personality is the central mechanism of the Moore plane. It’s a hybrid world: a familiar Euclidean interior stitched to a bizarrely discrete boundary. Let’s see what sort of universe this creates. Can we do geometry in it? Can we measure things? Does it behave in ways we might call "reasonable"?

Given the strange nature of the boundary, you might expect the Moore plane to be a complete chaotic mess. But, surprisingly, it passes several important tests for "good behavior."

First, it is a ​​Hausdorff​​ (or $T_2$) space. This is a very basic notion of sanity for a topological space. It simply means that for any two distinct points, say $p$ and $q$, you can find two disjoint open sets, one containing $p$ and the other containing $q$. You can put a "wall" between them. This is easy to see:

• If both points are inland ($p, q \in U$), we are in the Euclidean plane, and we can always draw two non-overlapping disks around them.
• If one is inland ($p \in U$) and one is on the coast ($q \in L$), we can draw a small disk around $p$ that stays far from the coast, and a sufficiently small tangent bubble at $q$ that doesn't reach high enough to touch the disk.
• If both are on the coast ($p, q \in L$), we just give them each a tangent bubble. If the bubbles are small enough, their horizontal separation will ensure they don't overlap.

So far, so good. The space is not a jumble; we can tell points apart. In fact, it's even better behaved. The Moore plane is a ​​regular​​ space. This is a step up from being Hausdorff. It means that if you have a closed set $C$ and a point $p$ not in $C$, you can separate them with disjoint open sets—you can put an open "moat" around the point and a separate open "fence" around the entire closed set. This property is crucial because it implies that the topology is rich enough to be described by continuous functions, making the space what topologists call a ​​Tychonoff space​​.

At this stage, the Moore plane looks quite respectable. It separates points, it separates points from closed sets... it seems to be climbing the ladder of "nice" topological spaces. It feels like the next logical step, normality, should hold as well.

Here is where our seemingly well-behaved world reveals a deep, fundamental flaw. A space is ​​normal​​ if it can separate not just a point and a closed set, but any two disjoint closed sets. It's the next rung on the ladder, and the Moore plane stumbles and falls right here.

To see this, we need to find two disjoint closed sets that defy all attempts to place them in separate open containers. The culprits live on the strange boundary line, $L$. Consider two sets:

• $A = \{(x, 0) \mid x \in \mathbb{Q}\}$, the set of all points on the $x$-axis with a rational coordinate.
• $B = \{(x, 0) \mid x \in \mathbb{R} \setminus \mathbb{Q}\}$, the set of all points on the $x$-axis with an irrational coordinate.

These two sets are disjoint, and as we saw, any subset of the x-axis is a closed set in the Moore plane's topology. So we have our test case: two disjoint closed sets, $A$ and $B$. Now, let's try to separate them. Suppose we could. This would mean there's an open set $U$ containing all the rational points $A$, and a disjoint open set $V$ containing all the irrational points $B$.

What would this imply? For every rational point $q \in A$, its open set $U$ must contain a little tangent bubble of some radius $r_q > 0$. For every irrational point $s \in B$, its open set $V$ must contain a tangent bubble of some radius $r_s > 0$. For $U$ and $V$ to be disjoint, all these bubbles for rationals must be disjoint from all the bubbles for irrationals.

There's a beautiful way to see why this is impossible, appealing to the idea of continuous functions (as formalized by Urysohn's Lemma). If $A$ and $B$ were separable, we could construct a continuous function $f: \Gamma \to [0, 1]$ that is $0$ on every point in $A$ and $1$ on every point in $B$. Now, think about what this function $f$ looks like just above the x-axis, say on the line $y=\epsilon$ for some tiny $\epsilon > 0$. The function values $f(x, \epsilon)$ must smoothly change as you vary $x$. But as you bring this line down toward the axis (as $\epsilon \to 0$), the values $f(x, \epsilon)$ must approach $0$ if $x$ is rational and $1$ if $x$ is irrational. In the limit, our well-behaved continuous function would have to morph into the pathological Dirichlet function, which is famously discontinuous everywhere. The pointwise limit of a sequence of continuous functions can't be this badly behaved! This contradiction tells us that our initial assumption must be wrong. No such continuous function can exist, and therefore, the sets $A$ and $B$ cannot be separated. The Moore plane is ​​not normal​​.

The geometric intuition is just as powerful. The rational and irrational numbers are so thoroughly intermingled that you can always find a rational point $q$ and an irrational point $s$ arbitrarily close to each other. No matter what radii $r_q$ and $r_s$ you choose for their bubbles, if you pick $q$ and $s$ close enough, the geometry of tangent circles dictates that their bubbles must inevitably intersect in the plane above. The attempt to build a wall between them is doomed to fail.

The failure of normality is not a minor blemish; it's a fatal flaw that brings down a whole house of cards of desirable properties.

First, the Moore plane is ​​not paracompact​​. Paracompactness is a more subtle notion of "niceness" related to how open covers of a space can be refined. For our purposes, we only need to know one crucial theorem: every paracompact Hausdorff space is also normal. We already know the Moore plane is Hausdorff but is not normal. The conclusion is immediate: it cannot be paracompact.

More devastatingly, the Moore plane is ​​not metrizable​​. This means there is no possible distance function $d(p_1, p_2)$ that can give rise to this specific topology. You simply cannot define a notion of "distance" that produces these tangent bubble neighborhoods. There are several ways to see why.

One elegant argument involves another one of the plane's properties: it is ​​separable​​, meaning it has a countable dense subset. The set of all points $(x, y)$ in the upper half-plane where both $x$ and $y$ are rational numbers is countable, and any open set in the Moore plane (whether a disk or a tangent bubble) will contain at least one of these points. So we have this countable set that gets "close" to everything. Now, contrast this with the x-axis, $L$. We saw that $L$ is an uncountable set where the subspace topology is discrete. A separable metrizable space simply cannot contain an uncountable closed discrete subspace. It's like trying to guard an infinite line of soldiers, each standing miles from the next, with only a finite number of guards. You can't do it.

The deepest reason, however, lies in the very structure of the open sets. The celebrated ​​Nagata-Smirnov Metrization Theorem​​ gives a precise condition for metrizability: a space is metrizable if and only if it is regular, Hausdorff, and has a "$\sigma$-locally finite basis". This last condition is a technical but crucial requirement on the "building blocks" (the basis) of the topology. It demands that the entire collection of basis sets can be organized into a countable number of "well-behaved" sub-collections. The Moore plane, despite being regular and Hausdorff, fails this final test. The uncountable, demanding nature of the boundary line $L$ makes it impossible to construct such a well-organized basis. Any attempt gets overwhelmed; a rigorous argument shows that there will always be some point on the line where this "local finiteness" condition breaks down spectacularly.

And so, the Moore plane stands as a masterful counterexample. It is deceptively simple to define, yet it lives in a subtle sweet spot: orderly enough to be regular and Hausdorff, but too wild on its boundary to be normal, paracompact, or metrizable. It is a stark and beautiful reminder that in the world of topology, our Euclidean intuitions can be a treacherous guide, and the simplest-looking rules can generate worlds of unexpected complexity.

Now that we have acquainted ourselves with the curious rules of the Moore plane, we are like explorers who have just learned the strange local customs of a newly discovered land. The real adventure begins when we start to walk around, to interact with this world, and to see how these customs lead to a society of shapes and forms that is both bizarre and profoundly instructive. The Moore plane is not merely a cabinet of curiosities; it is a laboratory where our deepest intuitions about space are put to the test. By observing where our familiar notions bend and break, we gain a much clearer understanding of the hidden machinery that underpins all of mathematics.

Let's begin our exploration on the boundary, the $x$-axis, where all the topological mischief originates. Imagine two points on this line, say at $-a$ and $a$. We decide to place a "protective bubble"—a basic neighborhood—around each one. The neighborhood at $-a$ is an open disk of radius $R$ tangent to the axis from above. Now, we ask: how large can we make the corresponding bubble at $a$ before it touches the first one?

Our Euclidean intuition might suggest a simple linear relationship. But the geometry of the Moore plane is more subtle and beautiful. The maximum radius $r$ we can give to the second bubble turns out to be $r_{\text{max}} = \frac{a^2}{R}$. This little formula is quite revealing! As the first bubble grows (larger $R$), the second one must shrink dramatically to stay disjoint. This inverse relationship tells us that neighborhoods on the boundary exert a powerful, long-range influence on one another, an effect not seen in ordinary Euclidean space.

What happens if we try to place an infinite number of these bubbles? Suppose we want to give a bubble of the same radius $r$ to every integer point on the $x$-axis. A simple calculation, driven by the geometry of tangent circles, shows that this is only possible if the radius $r$ is no larger than $\frac{1}{2}$. If we try to make the bubbles any bigger, they are forced to overlap. It feels as though the points on the boundary are engaged in a delicate, crowded dance, where everyone's personal space is strictly limited by their neighbors.

This "crowding" leads to an even more startling conclusion. Suppose we have two disjoint neighborhoods on the axis. In our familiar world, we would expect that their closures—the sets including their boundaries—would also be a certain distance apart. But in the Moore plane, this is not so. It is possible to choose two disjoint open neighborhoods whose closures become arbitrarily close, to the point where they actually touch. The infimum of the Euclidean distance between their closures is exactly zero! This means there's no "breathing room" between sets in the way we've come to expect. This single fact is a crack in the foundation of our intuition, a crack that will soon widen to reveal the plane's most famous properties.

Topologists love a good counterexample. A counterexample is a spotlight that illuminates the precise boundary of a theorem, showing us exactly how far a concept can be pushed before it fails. The Moore plane is perhaps the most fertile ground in all of topology for cultivating such examples.

Its most celebrated role is as a space that is regular but not normal. In simple terms, a normal space is one where any two disjoint closed sets can be separated by disjoint open "sleeves." We saw that the Moore plane struggles to keep sets apart; it turns out that for some sets, it fails completely. Consider the $x$-axis. The set of points with rational coordinates, $A = \{(x,0) \mid x \in \mathbb{Q}\}$, and the set of points with irrational coordinates, $B = \{(x,0) \mid x \in \mathbb{R} \setminus \mathbb{Q}\}$, are both closed sets in the Moore plane, and they are certainly disjoint. Yet, it is impossible to find two disjoint open sets $U$ and $V$ such that $A \subseteq U$ and $B \subseteq V$. Any attempt to "thicken" the set of rationals into an open set inevitably ends up "touching" the irrationals. The plane fails the fundamental test of normality.

This failure has cascading consequences. For instance, we often assume that nice properties of a space are inherited by its subspaces—a property called being "hereditary." The Moore plane demonstrates that this is not always true for separability. A space is separable if it contains a countable set of points that is "everywhere," like the rational numbers in the real line. The Moore plane itself is separable; the set of points with rational coordinates in the upper half-plane works just fine. But if we look at the subspace consisting only of the $x$-axis, something amazing happens. The topology it inherits is the discrete topology, where every single point is its own isolated open set! Since the real numbers are uncountable, this subspace is an uncountable collection of isolated points. It cannot possibly have a countable dense subset, and so it is not separable. This shows that a separable Moore space is not necessarily hereditarily separable, providing a crucial counterexample in the theory of generalized metric spaces. The strangeness of the Cantor set on this axis provides a similar insight, where an uncountable, perfect set in the standard topology becomes an uncountable, discrete, and non-compact set in the Moore plane's subspace topology.

Yet, for all its pathologies, the Moore plane is not entirely misbehaved. At every single one of its points—whether in the placid upper half-plane or on the tumultuous boundary—it is first-countable. This means that at any point, we can find a countable sequence of ever-shrinking neighborhoods that "zero in" on it. This property ensures that its local behavior is still somewhat tame, for instance, by guaranteeing that its "tightness" is countable everywhere. The Moore plane thus occupies a fascinating middle ground: locally manageable, but globally pathological.

The consequences of the Moore plane's peculiar structure ripple far beyond the borders of general topology, influencing fields like analysis and even algebraic topology.

In analysis, we are often concerned with the limits of functions and the paths curves can take. The Moore plane provides a sharp illustration of how topology governs this behavior. Consider a family of wildly oscillating functions, like $f_a(x) = x^a \sin(1/x)$, as $x$ approaches zero. Can the graph of such a function "approach" the origin $(0,0)$ in the Moore plane? The answer depends critically on the parameter $a$. It turns out that the origin is a limit point of the graph if and only if $a 2$. The geometric reason is beautiful: for the graph to enter a tangent-disk neighborhood of the origin, the function's value $y$ must be large enough relative to $x$ to satisfy $x^2+y^2 2ry$. The condition $a 2$ is precisely what allows the function's oscillations to "pierce" these bubbles, no matter how small they are. This provides a tangible link between an analytic property of a function ($a 2$) and a purely topological feature of the space.

Perhaps the most dramatic interdisciplinary connection lies in algebraic topology. Tools like the Mayer-Vietoris sequence are workhorses for calculating homology groups, which classify the "holes" in a space. The standard proof that this tool works requires the space to be normal. And where does that proof fail? On spaces like the Moore plane! The non-existence of a continuous function that separates the closed sets of rational and irrational points on the axis is not just a violation of Urysohn's Lemma; it's a fundamental obstruction that prevents the construction of a "partition of unity," a key ingredient in the proof. This tells us that the assumption of normality in a major algebraic theorem is not a mere technicality to be glossed over. The Moore plane stands as a stark warning sign, showing us the exact cliff edge where our powerful machinery can fail.

This intricate web of connections extends even to the abstract theory of metrization. We know the Moore plane is not metrizable—you can't define a distance function that gives you back its topology. Why not? The deep answer is that it is not paracompact. Paracompactness is a subtle "global tidiness" condition on the open covers of a space. For a Moore space, being metrizable and being paracompact are one and the same. The failure of the Moore plane to be normal is intrinsically linked to its failure to be paracompact. A certain construction aimed at "stratifying" the space fails because we cannot guarantee that we can shrink a "sleeve" around a closed set to be contained within a given larger open set, a property that is guaranteed in a paracompact space.

In the end, the Moore plane teaches us a lesson in humility and wonder. It shows that even in a seemingly simple setting like the upper half of a sheet of paper, a slight change in the rules of "nearness" can create a universe of unexpected complexity. It is by studying these frontiers, these places where our intuition fails, that we truly appreciate the depth, richness, and profound unity of mathematics.