Sorgenfrey Line

In the world of mathematics, our intuition is often shaped by the familiar landscape of the Euclidean real line, a seamless continuum where concepts like distance and movement are straightforward. But what happens if we slightly alter the fundamental rules that define this space? What if we change the very definition of an "open set"? The Sorgenfrey line emerges from this simple question, using half-open intervals instead of open ones. This seemingly minor tweak creates a topological space that is both bizarre and profoundly insightful, challenging our core assumptions about geometry and continuity. This article serves as a guide to this strange new world. In the following chapters, you will delve into the foundational properties of the Sorgenfrey line and its counter-intuitive consequences for connectedness and convergence. Following that, you will see how these principles are applied in the Sorgenfrey plane, a space that has become a famous "laboratory" for testing topological theories and generating crucial counterexamples that have shaped the field of general topology.

Imagine the familiar number line, a continuous, unbroken thread of points. We can glide smoothly along it, and our intuition about concepts like "nearness," "connectedness," and "limits" is built on this smooth world. Now, let's play a game. We're going to change just one tiny rule about what it means for a collection of points to be a "basic open set"—the fundamental building blocks of our space. Instead of using open intervals $(a, b)$ as our blocks, we will use half-open intervals of the form $[a, b)$. This new space, built from the same set of real numbers but with a different notion of openness, is the ​​Sorgenfrey line​​. It sounds like a minor tweak, but as we are about to see, this single change shatters our comfortable geometric world and rebuilds it into something utterly strange and beautiful.

What does this change from $(a,b)$ to $[a,b)$ really do? It creates a topology that is ​​finer​​, or more discerning, than the usual one. Think of it like upgrading a microscope; you can now see features that were previously blurred together. Every open set from the old, usual topology is still open here. For instance, an interval like $(0, 5)$ is still open because we can describe it as a union of our new building blocks, for example, the union of all sets $[x, 5)$ for every $x$ in $(0, 5)$.

However, the reverse is not true. Our new building blocks, the sets like $[0, 1)$, are open by definition in the Sorgenfrey line, but they are emphatically not open in the usual number line. Why? Because the point $0$ is included, but no matter how tiny an open interval you draw around $0$ in the usual sense, like $(-\epsilon, \epsilon)$, it will always include negative numbers that aren't in $[0, 1)$. This asymmetry gives us a powerful insight: the identity map that takes each point $x$ on the usual line to the same point $x$ on the Sorgenfrey line is an ​​open map​​ (it preserves old open sets) but is ​​not continuous​​ (it doesn't preserve the new ones when going backward). We have more open sets at our disposal, making the Sorgenfrey line a richer, more detailed landscape.

This new level of detail has a shocking consequence. Let's consider the interval $[0, 1)$. In the Sorgenfrey line, this set is open by definition—it's one of our basic building blocks. Now let's look at its complement, the set of all numbers not in $[0, 1)$, which is $(-\infty, 0) \cup [1, \infty)$. Is this complement open? Yes! Any point less than $0$ can be put in a little Sorgenfrey interval that is also less than $0$. Any point greater than or equal to $1$ can be the start of a new Sorgenfrey interval that stays above $1$. Since the complement is open, the original set, $[0, 1)$, must be ​​closed​​.

Pause and think about that. The set $[0, 1)$ is both open and closed at the same time. Such a set is called ​​clopen​​. In the familiar real line, the only clopen sets are the trivial ones: the empty set and the entire line itself. The existence of a non-trivial clopen set like $[0,1)$ is a topological bombshell. It immediately proves that the Sorgenfrey line is ​​not connected​​. It can be "separated" into the disjoint, non-empty open sets $[0,1)$ and its complement.

We can take this even further. It turns out that any subset of the Sorgenfrey line containing more than one point is disconnected. The only connected subsets are the individual points themselves. This means it's impossible to draw a continuous path from one point to another. A path, by definition, is the continuous image of a connected interval like $[0,1]$, and since continuous maps preserve connectedness, the image must be connected. But the only connected options are single points! Thus, any "path" on the Sorgenfrey line must be a constant function—it starts at a point and never leaves. The space is ​​totally path-disconnected​​. Every point is a lonely island, forever isolated from its neighbors. This also means the space is ​​not locally connected​​, as no point has a connected neighborhood other than the point itself, which isn't a neighborhood.

If you can't travel between points, what does it mean to "approach" a point? This brings us to the convergence of sequences. Consider the sequence $x_n = -1/n$: $-1, -1/2, -1/3, \ldots$. In the standard real line, this sequence famously converges to $0$. But what happens on the Sorgenfrey line?

Let's suppose it does converge to $0$. For this to be true, the sequence must eventually enter any open neighborhood of $0$ and stay there. One of the basic open neighborhoods of $0$ is the set $[0, 0.1)$. Now, look at our sequence. Every single term is negative. Not one of them is in the interval $[0, 0.1)$. The sequence approaches $0$ from the "wrong side"—a side that is topologically walled off. Therefore, the sequence does not converge to $0$. A similar argument shows it can't converge to any other point either.

This simple example reveals a deep truth: convergence on the Sorgenfrey line is much harder to achieve. A sequence $(x_n)$ converging to a point $L$ must satisfy $L \le x_n$ for all sufficiently large $n$. In other words, it must approach its limit from the right (or by being equal to it). A strictly increasing sequence, like our $x_n = -1/n$, can never converge! Because this sequence has no convergent subsequence, we can also conclude that the Sorgenfrey line is ​​not sequentially compact​​.

With its disconnectedness and strange convergence, the Sorgenfrey line might seem like a chaotic mess. But in other ways, it is surprisingly well-behaved. It adheres to important principles of separation. For instance, it is a ​​Hausdorff space​​. This means that for any two distinct points, say $x$ and $y$, we can always find two disjoint open "bubbles" that contain them. If $x y$, the open sets $U = [x, y)$ and $V = [y, y+1)$ do the job perfectly; $x$ is in $U$, $y$ is in $V$, and they don't overlap. This ensures that points are topologically distinguishable.

It's even better than that. The Sorgenfrey line is a ​​normal space​​ (and since it's also $T_1$, it's a ​​$T_4$ space​​). This means we can separate not just two points, but any two disjoint closed sets. Imagine two complicated, infinite, and intertwined closed sets $A$ and $B$. As long as they don't share any points, we can find two large, disjoint open sets $U$ and $V$ such that $A$ is completely inside $U$ and $B$ is completely inside $V$. This is a high degree of "separation civility," showing that despite its pathological disconnectedness, the space has a very orderly structure.

As we conclude our initial exploration, we find the Sorgenfrey line to be a place of contrasts. Some features are surprisingly familiar, while others hint at even deeper strangeness.

For one, the set of rational numbers, $\mathbb{Q}$, is still ​​dense​​ here. Any basic open set $[a,b)$ contains the standard open interval $(a,b)$, and we know from our old world that such an interval must contain a rational number. So, no matter how small our Sorgenfrey building block, it's guaranteed to have a rational point inside. The intricate weave of rational and irrational numbers is preserved.

But what happens if we use this line to build a plane? We create the ​​Sorgenfrey plane​​, $\mathbb{R}_S \times \mathbb{R}_S$, where the basic open sets are rectangles of the form $[a, b) \times [c, d)$. Here, intuition fails spectacularly. Consider the "open" unit square $A = (0, 1) \times (0, 1)$. What is its boundary? In the standard plane, it's the four-sided frame. But here, the boundary is only the bottom edge $\{ (x, 0) \mid 0 \le x 1 \}$ and the left edge $\{ (0, y) \mid 0 \le y 1 \}$. The top and right edges are "pushed away" by the topology. This strange geometry is a direct consequence of our half-open building blocks. And it is here, in this product space, that one of topology's most famous counterexamples lives: the Sorgenfrey line is normal, but the Sorgenfrey plane is not. This single, simple change to our definition of an interval has given us a space that is both orderly and wild, a perfect laboratory for testing the limits of our geometric intuition.

Having grasped the foundational principles of the Sorgenfrey line, we are now like explorers who have learned the peculiar grammar of a new land. Our next adventure is to venture into this territory and see what structures it builds, what familiar landscapes it transforms, and what secrets it reveals about the universe of topology itself. We will focus our journey on the ​​Sorgenfrey plane​​, the product space $\mathbb{R}_S \times \mathbb{R}_S$, a world constructed from two Sorgenfrey lines. You will soon discover that this plane is no mere curiosity; it is a master laboratory for testing the limits of our mathematical intuition and a source of profound insights that have shaped the field of general topology.

Our first stop is a close examination of the Sorgenfrey plane's local geography. In the familiar Euclidean plane, the "boundary" of an open rectangle is its perimeter. What about here? The basic building blocks of the Sorgenfrey plane are rectangles of the form $S = [a, b) \times [c, d)$. As we discovered in the previous chapter, the interval $[a, b)$ is not just open in $\mathbb{R}_S$; it is also closed. Since the Sorgenfrey plane is a product of these spaces, the basic rectangle $S$ is also both open and closed—it is "clopen." This has a startling consequence: the boundary of such a set, defined as its closure minus its interior, is simply the set minus itself. The boundary is empty!. Imagine a country with no borders; this is the nature of the fundamental regions in the Sorgenfrey plane. This property immediately signals that we are in a space that is profoundly "disconnected" at a microscopic level.

Now, let's see what happens to familiar geometric figures, like straight lines.

A horizontal line, say $y = c_0$, is a subspace of the plane. What topology does it inherit? If we intersect our basic open rectangles $[a, b) \times [c, d)$ with this line, we only get non-empty sets if $c_0$ is in $[c, d)$. The resulting intersection is essentially the set of points $(x, c_0)$ where $x \in [a, b)$. This reveals that the inherited topology on the horizontal line is none other than the Sorgenfrey line topology itself. The same logic applies to vertical lines and, with a little more work, to any line with a positive slope. These lines, embedded in the Sorgenfrey plane, are just copies of the Sorgenfrey line.

But what about a line with a negative slope? Here, our Euclidean intuition shatters completely. Consider the "anti-diagonal," the line $L = \{(x, -x) \mid x \in \mathbb{R}\}$. Let's pick a point on this line, say $p = (2, -2)$. Can we find an open neighborhood of $p$ that contains no other points of the line $L$? In the Euclidean world, this is impossible. But in the Sorgenfrey plane, we can. The open set $N = [2, 3) \times [-2, -1)$ does the trick. For any point $(x, -x)$ to be in $N$, we need $x \in [2, 3)$ and $-x \in [-2, -1)$. The second condition means $1 x \le 2$. The only value of $x$ satisfying both conditions is $x=2$. So, the neighborhood $N$ intersects the line $L$ only at the point $p$.

This is remarkable! Every single point on the anti-diagonal can be isolated from its brethren with an open set. This means the subspace topology on the anti-diagonal is the ​​discrete topology​​—every point is an open set unto itself. The line, which we think of as the epitome of continuity, has been pulverized into a collection of isolated points.

This strange, "pathological" behavior is not just a party trick; it is the Sorgenfrey plane's most important feature. Topologists are like physicists who propose general laws and then search for experiments to confirm or deny them. The Sorgenfrey plane is a powerful apparatus for running these "experiments" on topological conjectures. Many seemingly reasonable ideas have met their end in this plane.

A fundamental property of a space is ​​connectedness​​. Can the space be broken into two disjoint, non-empty open pieces? The Sorgenfrey plane, riddled with clopen sets, is a prime suspect for being disconnected. And indeed it is. The sets $U = \{(x,y) \mid x+y 0\}$ and $V = \{(x,y) \mid x+y \ge 0\}$ form a perfect separation. Both are non-empty, disjoint, their union is the whole plane, and, crucially, both can be shown to be open in the Sorgenfrey topology. The plane cleanly splits in two.

The most famous applications, however, involve properties that are more subtle. Consider these two plausible-sounding statements:

1. The product of two separable spaces is separable. (A space is ​​separable​​ if it contains a countable subset that is "everywhere," i.e., dense).
2. The product of two Lindelöf spaces is Lindelöf. (A space is ​​Lindelöf​​ if from any collection of open sets that covers the space, you can pick a countable number of them that still cover it).

The Sorgenfrey line $\mathbb{R}_S$ itself is "well-behaved" in these respects. It is separable, as the set of rational numbers $\mathbb{Q}$ is still dense. It is also a Lindelöf space. So, what about its product, the Sorgenfrey plane $\mathbb{R}_S \times \mathbb{R}_S$?

Here is where the pulverized anti-diagonal returns to do its devastating work. We saw that the anti-diagonal is an uncountable set with the discrete topology. We can also show it is a ​​closed​​ subset of the plane,.

• ​​Failure of Separability:​​ A separable space cannot contain an uncountable discrete subspace. If it did, each of the uncountably many points would need to be "touched" by our countable dense set, which is impossible since the points are isolated from each other. Since the Sorgenfrey plane contains the uncountable, discrete anti-diagonal, it cannot be separable. Thus, the product of two separable spaces is not always separable.

• ​​Failure of the Lindelöf Property:​​ A similar argument works for the Lindelöf property. An uncountable discrete space cannot be Lindelöf; the open cover consisting of all the individual points has no countable subcover. Since being Lindelöf is a property that is inherited by closed subspaces, if the Sorgenfrey plane were Lindelöf, its closed subspace, the anti-diagonal, would have to be Lindelöf too. But it isn't. Therefore, the Sorgenfrey plane is not Lindelöf.

The Sorgenfrey plane stands as a monumental counterexample, teaching us that these important topological properties are not always preserved by products. It forces us to be more careful, to refine our theorems, and to appreciate the deep subtleties of topology.

The Sorgenfrey plane is more than just a destroyer of conjectures. Its unique structure allows for a fine-grained analysis of geometric and topological properties. Let's return to the question of lines in the plane. We can ask: which of these lines are ​​metrizable​​, meaning their topology can be described by some distance function?

The answer is a beautiful synthesis of everything we have learned.

• Lines with a negative slope, as we saw, are discrete. Any discrete space is metrizable (for instance, by a metric where the distance is $1$ between distinct points and $0$ otherwise). So, these lines are metrizable.
• Lines with non-negative slope (horizontal, vertical, or positive slope) are all homeomorphic to the Sorgenfrey line $\mathbb{R}_S$. But is $\mathbb{R}_S$ itself metrizable? No! A key theorem states that for a metric space, separability is equivalent to being second-countable (having a countable basis). The Sorgenfrey line is separable but not second-countable, so it cannot be metrizable.

Therefore, the complete classification is: a line in the Sorgenfrey plane is metrizable if and only if it has a negative slope. This is not just a curious fact; it's a demonstration of how topology provides precise tools to classify and understand geometric objects in ways that go far beyond simple visual appearance.

The ideas can be extended even further. What if we build a hybrid space, taking one standard real line and one Sorgenfrey line, to form the product $\mathbb{R} \times \mathbb{R}_S$? What happens to the anti-diagonal here? A similar analysis reveals that it is no longer discrete, nor is it the Sorgenfrey line. Instead, it becomes homeomorphic to the ​​upper limit topology​​ on $\mathbb{R}$, whose basic sets are of the form $(a, b]$. This elegant result shows how these different topologies are related, like members of a family, and how combining them in product spaces can generate new and interesting structures.

In the end, the Sorgenfrey plane might seem "pathological" from the comfortable viewpoint of Euclidean geometry. But in science, it is often the exceptions, the pathologies, that teach us the most. They reveal the hidden assumptions in our theories and force us to build a richer, more robust understanding of the world. The Sorgenfrey plane is a beautiful monster, and by studying its anatomy, we learn the true meaning of the topological concepts that govern all spaces, from the simplest to the most complex.