Subbasis in Topology

In the mathematical field of topology, we study the properties of spaces that are preserved under continuous deformations. These spaces are defined by collections of "open sets," which can be overwhelmingly complex. To manage this complexity, mathematicians use a simpler collection of sets called a basis, from which all open sets can be constructed. But can we go deeper? Is there an even more fundamental set of building blocks from which the basis itself is formed? This question leads us to the concept of a ​​subbasis​​, the "atomic" ingredients of a topological space. This article explores how these elementary sets provide a powerful and elegant framework for both understanding and constructing complex topological structures.

This article is divided into two chapters. First, in "Principles and Mechanisms," we will explore the formal mechanics of a subbasis, detailing the two-step recipe—intersect, then unite—that allows us to generate an entire topology from a simple collection of sets. Then, in "Applications and Interdisciplinary Connections," we will uncover the true power of this concept, from its role as a "master key" in proving one of topology's most important properties, compactness, to its function as an architect's blueprint for defining topologies on abstract realms like function spaces and even the spaces of mathematical logic.

Imagine you want to describe a city. You could try to list every single possible shape of land you could own—an impossible task. Or, you could define a set of basic, rectangular city blocks. Then, any property you could own would be a combination of these blocks. This is the idea behind a ​​basis​​ in topology. The "open sets"—the fundamental regions of a topological space—are all the possible combinations (unions) of these basic "blocks".

But can we go deeper? Where do the standard blocks come from? Can we build them from something even more elementary? What if we only had a set of straight lines (representing, say, property boundaries)? We couldn't form a block with a single line, but by taking two horizontal and two vertical lines, their intersection defines a rectangular block. This is the leap from a basis to a ​​subbasis​​. A subbasis is a collection of sets that might be too simple on their own, but through the process of intersection, they generate the more useful "blocks" of our basis. The subbasis represents the most fundamental "atomic" ingredients of a topological space.

The journey from a handful of atomic subbasis sets to a full-fledged topology is a beautiful and simple two-step process.

First, you ​​generate a basis by taking finite intersections​​. You take your subbasis sets and find all the regions that are common to any finite collection of them. These intersections are the "bricks" or "city blocks" of your space—they form the basis.

Second, you ​​generate the topology by taking arbitrary unions​​. Once you have your full set of bricks (the basis), you can construct every possible "property" (open set) by gluing any number of them together.

Let's see this in action with a simple universe containing just four points, $X = \{1, 2, 3, 4\}$. Suppose we declare our atomic sets—our subbasis—to be $\mathcal{S} = \{\{1,2\}, \{2,3\}, \{3,4\}\}$. Let's build the topology.

1. ​​Step 1: Make the bricks (the basis $\mathcal{B}$)​​. We take all finite intersections of sets in $\mathcal{S}$:

   • The sets themselves: $\{1,2\}$, $\{2,3\}$, $\{3,4\}$.
   • Intersections of two sets: $\{1,2\} \cap \{2,3\} = \{2\}$, and $\{2,3\} \cap \{3,4\} = \{3\}$. (Note that $\{1,2\} \cap \{3,4\}$ is the empty set, $\emptyset$).
   • Intersection of all three: $\{1,2\} \cap \{2,3\} \cap \{3,4\} = \emptyset$.
   • By convention, the "empty" intersection (of zero sets) gives us the whole universe, $X$. So, our collection of bricks is $\mathcal{B} = \{\emptyset, \{2\}, \{3\}, \{1,2\}, \{2,3\}, \{3,4\}, X\}$.

2. ​​Step 2: Build the rooms (the topology $\mathcal{T}$)​​. We take all possible unions of our bricks from $\mathcal{B}$:

   • The bricks themselves are open sets.
   • We can also form new sets like $\{1,2\} \cup \{3\} = \{1,2,3\}$ and $\{2\} \cup \{3,4\} = \{2,3,4\}$.
   • All told, we find exactly nine distinct open sets: $\mathcal{T} = \{\emptyset, \{2\}, \{3\}, \{1,2\}, \{2,3\}, \{3,4\}, \{1,2,3\}, \{2,3,4\}, X\}$.

And there we have it. From just three simple sets in our subbasis, a specific and non-trivial topological structure emerges. Sometimes this process is even simpler. If we start with the subbasis on the integers $\mathbb{Z}$ consisting of all downward-pointing rays $\mathcal{S} = \{\dots, \{n \mid n \le -1\}, \{n \mid n \le 0\}, \{n \mid n \le 1\}, \dots\}$, the intersection of any two rays, say $\{n \mid n \le k_1\}$ and $\{n \mid n \le k_2\}$, is just the smaller of the two rays. The basis generated is just the subbasis itself (plus $\mathbb{Z}$), which creates a rather curious topology where the only "standard" open sets are these rays.

You might ask: why bother with the extra step? Why not just define the basis directly? The answer is elegance and efficiency. A subbasis is often vastly simpler than the basis it generates.

Consider the ​​discrete topology​​, where every subset is an open set. On our four-point set $X = \{1,2,3,4\}$, this means there are $2^4 = 16$ open sets. The most obvious basis is the collection of single-point sets: $\mathcal{B} = \{\{1\}, \{2\}, \{3\}, \{4\}\}$, since any set can be built by uniting single points. But can we find a subbasis for this topology that doesn't contain any of these single-point sets?

It seems paradoxical, but the answer is yes! Consider the subbasis $\mathcal{S} = \{\{2,3,4\}, \{1,3,4\}, \{1,2,4\}, \{1,2,3\}\}$. None of these sets have size one. But watch what happens when we intersect them: $\{1,3,4\} \cap \{1,2,4\} \cap \{1,2,3\} = \{1\}$ By intersecting the three sets that contain 1, we isolate 1 itself! We have "carved out" the singleton $\{1\}$ from larger sets. We can do this for every point, generating the basis of all singletons, which in turn generates all 16 open sets of the discrete topology. We've created the most fine-grained topology imaginable from a subbasis of large, coarse sets. This is the magic of the intersection rule.

This principle of economy has profound consequences. A topological space is called ​​second-countable​​ if it has a countable basis. This is a very powerful property that many important spaces like the real line $\mathbb{R}$ possess. The incredible fact is that if a space has a ​​countable subbasis​​, it is guaranteed to be second-countable. At first glance, this seems wrong. If we start with a countably infinite subbasis, won't the set of all finite intersections be uncontrollably large? No! The set of all pairs of elements is countable, the set of all triples is countable, and so on. The total collection of all finite intersections is a countable union of countable sets, which is itself countable. So, a simple countable description (the subbasis) is enough to guarantee a countable basis, which is a key ingredient for doing calculus and analysis. For most infinite spaces, the "size" of the smallest subbasis is the same as the "size" of the smallest basis.

Subbases give us a wonderful control panel for designing and comparing topologies. It's an intuitive principle that if you start with more raw materials, you can build more things. If we have two subbases $\mathcal{S}_1$ and $\mathcal{S}_2$ on the same set $X$, and $\mathcal{S}_1$ is a subset of $\mathcal{S}_2$, then every brick we can make from $\mathcal{S}_1$ is also a brick we can make from $\mathcal{S}_2$. This means the topology generated by $\mathcal{S}_1$ will be contained within the topology generated by $\mathcal{S}_2$. We say $\mathcal{T}_2$ is ​​finer​​ than $\mathcal{T}_1$, and $\mathcal{T}_1$ is ​​coarser​​ than $\mathcal{T}_2$. More subbasis elements lead to a richer, more detailed topology with more open sets.

However, the relationship isn't always so simple. It's not just the number of subbasis sets that matters, but their structure. Let's go back to a three-point universe $X=\{1,2,3\}$. Consider two different subbases:

• $\mathcal{S}_1 = \{\{1,2\}, \{2,3\}\}$
• $\mathcal{S}_2 = \{\{1\}, \{2,3\}\}$

Let's see what topologies they generate.

• From $\mathcal{S}_1$, we get the basis $\mathcal{B}_1 = \{X, \{1,2\}, \{2,3\}, \{2\}\}$. This generates the topology $\mathcal{T}_1 = \{\emptyset, \{2\}, \{1,2\}, \{2,3\}, X\}$. Notice the open set $\{2\}$.
• From $\mathcal{S}_2$, we get the basis $\mathcal{B}_2 = \{X, \{1\}, \{2,3\}\}$. This generates the topology $\mathcal{T}_2 = \{\emptyset, \{1\}, \{2,3\}, X\}$. Notice the open set $\{1\}$.

Now, compare them. $\mathcal{T}_1$ is not finer than $\mathcal{T}_2$ because $\mathcal{T}_1$ does not contain the set $\{1\}$. And $\mathcal{T}_2$ is not finer than $\mathcal{T}_1$ because $\mathcal{T}_2$ does not contain the set $\{2\}$. They are ​​incomparable​​. By swapping just one set in the subbasis—trading $\{1,2\}$ for $\{1\}$—we created a fundamentally different world, not just a more or less detailed version of the same one.

So far, we've treated subbases as a clever way to build familiar spaces. But their true value shines when we venture into territories where our geometric intuition fails. Consider the set of all continuous functions from a space $X$ to a space $Y$, denoted $C(X,Y)$. This is not a simple line or plane; it's a vast, infinite-dimensional "universe of functions". How can we define what it means for a set of functions to be "open"?

We can't draw a picture, but we can state a simple, intuitive requirement. We might say a collection of functions is a "neighborhood" if they all behave similarly at a particular point. This is the key idea behind the ​​topology of pointwise convergence​​. We define its subbasis as follows: for any point $x \in X$ and any open set $V$ in the target space $Y$, we define a subbasis element $S(x,V)$ to be the set of all continuous functions $f$ that map the point $x$ into the set $V$. $S(x,V) = \{ f \in C(X,Y) \mid f(x) \in V \}$ This is our "atomic" requirement. Now, we turn the crank. A basis element is a finite intersection of these sets, like $S(x_1, V_1) \cap S(x_2, V_2)$. This corresponds to the set of all functions $f$ that satisfy a finite list of conditions: $f(x_1)$ must be in $V_1$, $f(x_2)$ must be in $V_2$, and so on. An open set is then any union of these basis sets.

This is a breathtakingly elegant construction. The subbasis gives us a simple, powerful, and intuitive way to define a natural structure on an otherwise impossibly complex space. This very construction is a special case of a grand, unifying idea: the ​​initial topology​​. Given any set $X$ and any collection of functions mapping from $X$ to other topological spaces, the initial topology on $X$ is defined as the coarsest topology that makes all those functions continuous. And how is this topology generated? Its canonical subbasis is precisely the collection of all preimages of open sets under these functions. The subbasis is not just a convenient tool; it is the essential and minimal structure needed to talk about continuity. It is the genetic code that gives birth to the shape of space.

Now that we have met the subbasis and understand its formal mechanics from the previous chapter, you might be asking the perfectly reasonable question, "What is it good for?" Is it merely a bit of esoteric machinery for the professional topologist? Or does it, like so many great ideas in mathematics, offer us a new and powerful way to see the world? The answer, it turns out, is a resounding "yes" to the latter. The concept of a subbasis is not just a tool; it is a lens, a key, and a blueprint. It allows us to prove profound theorems with startling simplicity, to build new and essential mathematical worlds, and to uncover deep, unexpected connections between fields that, on the surface, seem to have nothing to do with each other.

Perhaps the most dramatic application of the subbasis is in unlocking the secrets of compactness. As we've seen, proving a space is compact directly from the definition—by showing that every possible open cover has a finite subcover—is a Herculean task. It's like trying to prove a building is waterproof by testing it against every imaginable configuration of rainfall. The Alexander Subbase Theorem hands us a master key. It tells us we don't need to check every storm; we just need to check a few elemental patterns of rain, those given by a subbasis. If the building holds up against those, it will hold up against any combination.

Let’s see this key in action. To get a feel for it, consider the most barren topological landscape imaginable: a set $X$ with the indiscrete topology, where the only open sets are $\emptyset$ and $X$ itself. Is this space compact? We can choose the simplest possible subbasis, the collection containing just the set $X$ itself, $\mathcal{S} = \{X\}$. Any cover of our space using sets from $\mathcal{S}$ must, well, use the set $X$. And the collection $\{X\}$ is a perfectly good finite subcover of size one! The theorem clicks, the lock turns, and we conclude the space is compact. It feels almost like a cheat, but it's a perfect illustration of the logic: the theorem's power lies in letting us choose the most convenient subbasis.

This becomes truly powerful when we visit more exotic spaces. Consider an infinite set $X$ with the cofinite topology, where a set is open if it's empty or its complement is finite. Is this strange space compact? Our intuition, forged in the Euclidean world, might stumble. But let's build a subbasis. A natural choice is the collection of all sets that are missing just a single point, $\mathcal{S} = \{X \setminus \{x\} \mid x \in X\}$. Now, let's take any cover of $X$ made from these subbasic sets. Pick any set from our cover, say $S_0 = X \setminus \{x_0\}$. This set covers almost everything, but it misses the point $x_0$. Since we have a cover, some other set in it, say $S_1 = X \setminus \{x_1\}$, must contain $x_0$. But this means $x_0 \neq x_1$. What happens when we unite these two sets? We get $(X \setminus \{x_0\}) \cup (X \setminus \{x_1\})$, which is all of $X$! Just two sets from our arbitrary subbasic cover are enough to cover the entire space. The master key has worked again, revealing the hidden compactness of this non-intuitive world with breathtaking simplicity.

The theorem doesn't just work for strange spaces; it can provide new and beautiful insights into familiar ones. We all learn in analysis that a closed interval $[a, b]$ in the real line is compact. The traditional proof can be quite involved. With Alexander's theorem, we can see it from a new angle. Let's take a subbasis consisting of all "open rays" starting from the left, like $[a, c)$, and all rays starting from the right, like $(d, b]$. Now, imagine any cover of $[a, b]$ built from these rays. If we are to cover the whole interval, the collection of left-rays must collectively stretch far enough to the right, and the collection of right-rays must stretch far enough to the left. More precisely, the "furthest" reach of any left-ray, let's call it $\sup(S_L)$, must extend beyond the "leftmost" starting point of any right-ray, $\inf(S_R)$. The moment $\sup(S_L) > \inf(S_R)$, we know there's a left-ray $[a, c)$ and a right-ray $(d, b]$ with $d < c$, and these two sets alone are sufficient to cover the entire interval $[a, b]$. The entire proof boils down to this elegant, intuitive argument about endpoints crossing over.

The true triumph of this method, however, comes when we start building more complex spaces. What about a product of two compact spaces, like the torus, which is just a product of two circles, $T^2 = S^1 \times S^1$? We can form a subbasis for the torus from the open sets of its constituent circles. The proof strategy then becomes wonderfully clear: if we assume there's a subbasic cover of the torus with no finite subcover, it must be because the open sets projected onto at least one of the circles fail to cover that circle. This generalizes. In what stands as one of the crown jewels of general topology, the method extends to any product of compact spaces, even an infinite one. Consider the Hilbert cube, $H = [0,1]^{\mathbb{N}}$, the space of all infinite sequences where each term is a number in $[0,1]$. This is an infinite-dimensional space where our geometric intuition falters. Yet, the Alexander Subbase Theorem handles it with ease. By assuming a subbasic cover has no finite subcover, we can perform a beautiful diagonal-style argument to construct a single "rogue point" in the Hilbert cube that, by its very construction, cannot be in any of the sets of the cover—a contradiction. This proof is the heart of Tychonoff's Theorem, which states that any product of compact spaces is compact, a result so fundamental that it is equivalent to the Axiom of Choice.

Of course, the key must also tell us when the door is locked. A subbasis can be used to prove a space is not compact. We simply need to find one cover made from subbasic sets that has no finite subcover. For the real line with the "right-ray topology" (where open sets are of the form $[a, \infty)$ and their unions), the collection of sets $\{[n, \infty) \mid n \in \mathbb{Z}\}$ covers the entire line, but no finite number of them can, as their union will always be bounded below. The same logic applies to other non-compact spaces like the Sorgenfrey line.

So far, we have used the subbasis as a tool for analysis, for investigating a pre-existing topology. But it has another, perhaps even more fundamental, role: that of a blueprint for construction. Often, we start with just a set and a desired notion of closeness, and we want to build a topology that captures it. A subbasis is the perfect way to do this. We simply state which sets we want, at a bare minimum, to be open, and the rules of topology automatically construct the entire, consistent structure for us from this simple blueprint.

This idea is nowhere more important than in the study of function spaces. Imagine the set of all possible configurations of a system of electronic switches, where each switch can be 'off', 'standby', or 'on'. Each configuration is a function from the set of switches to the set of states. What is a natural topology on this space of configurations? We can specify a subbasis. We can declare that the set of all configurations where "switch $s_1$ is on" is an open set. And the set where "switch $s_2$ is off" is an open set, and so on. The topology generated from this subbasis is the topology of pointwise convergence, where two configurations are "close" if they agree on the state of many individual switches. This is an incredibly useful and natural idea, and it's precisely the product topology we encountered with the torus and Hilbert cube.

This architectural approach allows us to define various topologies on the same set of functions, each suited for a different purpose. The compact-open topology, for instance, is another crucial structure on function spaces, generated by a different subbasic blueprint. It is designed to capture a more global sense of "closeness" between functions. It is a sign of the deep consistency of these ideas that in simple cases, such as functions on a finite set, this more sophisticated topology beautifully simplifies to become the same as the familiar product topology, reassuring us that our foundational concepts are sound.

The reach of topology is vast, and the concept of a subbasis provides a bridge to one of the most fundamental areas of human thought: mathematical logic. At first glance, what could geometry possibly have to do with the abstract rules of deduction?

Model theory, a branch of logic, studies the relationship between formal languages and mathematical structures. Within this field, one considers objects called "types." An $n$-type over a set of parameters $A$ can be thought of as a complete, consistent description of the properties an $n$-tuple of variables could possibly have, using formulas with parameters from $A$. The set of all such complete descriptions forms a space, called the Stone space, $S_n(A)$.

The question then arises: how should we structure this space of possibilities? What does it mean for two "complete descriptions" to be close to one another? The subbasis provides a stunningly elegant answer. We define a topology where the basic open sets correspond to single formulas. The set of all types that contain the formula $\varphi(\bar{x}, \bar{a})$ is declared to be an open set, denoted $[\varphi(\bar{x}, \bar{a})]$. This collection of sets, for all possible formulas, forms a basis (and thus, a subbasis) for the topology.

This isn't just a formal game. The space that emerges, the Stone space, has remarkable properties: it is compact, Hausdorff, and totally disconnected. And here is the punchline: the topological compactness of this space is a direct consequence and restatement of the Compactness Theorem of first-order logic, a pillar of modern logic which states that if every finite subset of a set of sentences has a model, then the entire set has a model. The abstract notion of topological compactness, born from the study of real numbers and geometry, provides a perfect geometric language for a fundamental principle of logical deduction.

From proving that a simple interval is compact, to defining the structure of infinite-dimensional spaces, and even to providing a geometric home for the very laws of logic, the subbasis reveals itself. It is a simple concept with profound consequences, a testament to the unifying beauty and power that lies at the heart of mathematics.