Continuity in Topology

In mathematics, the concept of a "continuous" function is fundamental, often first encountered in calculus as a function whose graph can be drawn without lifting the pen. This intuitive idea is formalized by the epsilon-delta definition, which is intrinsically tied to the notion of distance. But what happens in spaces where distance is meaningless? How can we describe a continuous transformation between abstract structures like legal arguments or color palettes? This article addresses this gap by introducing the more general and powerful topological definition of continuity, which replaces distance with the foundational concept of open sets. In the following chapters, we will first unravel the principles and mechanisms of this definition, exploring how it works through core examples. We will then journey through its diverse applications and interdisciplinary connections, discovering how topological continuity serves as a unifying thread connecting fields from analysis and algebra to geometry, revealing the deep structural skeleton of modern mathematics.

If you've studied calculus, you likely have a firm idea of what a continuous function is. You probably learned the famous epsilon-delta ($\epsilon-\delta$) definition, a wonderfully precise but notoriously tricky bit of machinery involving distances. It paints a picture of continuity like this: for any point, if you want the function's output to be very close to a certain value, you just need to make sure your input is close enough to the original point. It's a definition fundamentally tied to the idea of distance.

But what if you wanted to discuss continuity in a context where "distance" makes no sense? Can we talk about a "continuous" deformation of a space of all possible colors, or of all possible legal arguments? To do this, mathematicians developed a more profound and general idea, one that captures the intuitive essence of continuity—the idea of "not tearing"—without ever mentioning the word "distance." This is the topological definition of continuity, and it is built not on distance, but on the more primitive concept of ​​open sets​​.

An open set, in essence, is a region that doesn't include its own boundary. The interval $(0, 1)$ is open; the interval $[0, 1]$ is not. In topology, we define a space by simply declaring which of its subsets we will call "open." These collections of open sets, called ​​topologies​​, must follow a few simple rules (like the union of any open sets being open, and the intersection of a finite number of them being open), but beyond that, the possibilities are endless.

Here is the central principle:

A function $f$ from a space $X$ to a space $Y$ is ​​continuous​​ if for every open set $V$ in the destination space $Y$, its ​​preimage​​, $f^{-1}(V)$, is an open set in the starting space $X$.

Notice the direction! It's backward. We don't ask what the function does to open sets in its domain. Instead, we pick an open set in the codomain (the destination) and ask: where did it come from? The set of all points in the domain that map into this open set is called the preimage. If this preimage is always an open set in the domain, for every possible choice of open set in the codomain, then the function is continuous.

Why this backward-seeming definition? It beautifully guarantees that the function doesn't "tear" the space. If points that are close together in the domain were sent to points that are far apart in the codomain, we could find a small open "bubble" in the codomain that separates those output points. The preimage of this bubble would then be a disconnected, "torn" set in the domain. Our definition forbids this.

To get a feel for this definition, let's look at functions so simple that their continuity is almost a matter of logic.

First, consider the most basic function imaginable: the ​​identity function​​, $f(x) = x$, which maps a space $X$ to itself. If we take any open set $U$ in the codomain (which is just $X$), what is its preimage? The preimage is the set of all points $x$ in the domain such that $f(x)$ is in $U$. Since $f(x)=x$, this is just the set of all points $x$ such that $x$ is in $U$. In other words, the preimage of $U$ is $U$ itself!

Since we started by assuming $U$ was open, its preimage is automatically open. This holds for any open set $U$. So, the identity function is always continuous. It does nothing, so it certainly doesn't tear anything.

Now, let's try another universal case: the ​​constant function​​, $f(x) = c$ for all $x$ in a space $X$, where $c$ is a fixed point in a space $Y$. To check for continuity, we pick an arbitrary open set $V$ in $Y$. What is its preimage? There are only two possibilities:

1. If the point $c$ is inside $V$, then every point $x$ in $X$ gets mapped to $c$, which is in $V$. So, the preimage $f^{-1}(V)$ is the entire domain, $X$.
2. If the point $c$ is not inside $V$, then no point $x$ in $X$ gets mapped into $V$. So, the preimage $f^{-1}(V)$ is the empty set, $\emptyset$.

Here's the punchline: by the fundamental axioms of topology, the entire space $X$ and the empty set $\emptyset$ are always defined to be open sets, no matter what the topology on $X$ is! Therefore, the preimage of any open set is always open. The constant function, which collapses the entire universe to a single point, is paradoxically, and beautifully, always continuous.

These first examples might suggest that continuity is a simple property of a function's formula. But the real power and subtlety of the topological definition arise when we realize that continuity depends critically on the choice of topologies for the domain and codomain. A function's formula alone tells you nothing; you must know what is considered "open" in both spaces.

Let's explore the extremes. Suppose we give the domain $X$ the ​​discrete topology​​, where every single subset of $X$ is declared to be open. Now take any function $f: X \to Y$ to any space $Y$. Is it continuous? Let's check. Pick an open set $V$ in $Y$. Its preimage, $f^{-1}(V)$, is some subset of $X$. But in the discrete topology, all subsets are open! So the condition is automatically satisfied. Any function whose domain is a discrete space is continuous. The domain is so fundamentally "separated" into individual points that it's impossible to tear it.

What about the opposite extreme? Let the domain $X$ have the ​​indiscrete topology​​, where the only open sets are $\emptyset$ and the entire space $X$. For a function $f: X \to Y$ to be continuous, the preimage of any open set in $Y$ must be either $\emptyset$ or $X$. This is a severe restriction! Unless the codomain also has a very simple structure, very few functions will satisfy this.

These extremes teach us a vital lesson: having more open sets in the domain makes it easier for a function to be continuous. A topology with "more" open sets is called ​​finer​​, while one with "fewer" is called ​​coarser​​.

Let's see this in action with the identity function, $f(x)=x$, but on spaces with different topologies. Consider the real number line $\mathbb{R}$. The ​​standard topology​​ is the one you're used to, built from open intervals $(a, b)$. The ​​lower-limit topology​​ (or Sorgenfrey line) is built from half-open intervals $[a, b)$. Every standard open interval $(a, b)$ can be written as a union of lower-limit intervals (e.g., $(a,b) = \bigcup_{x \in (a,b)} [x, b)$), but not the other way around. Thus, the lower-limit topology is strictly finer than the standard one.

• ​​Case 1:​​ $f: (\mathbb{R}, \text{lower-limit}) \to (\mathbb{R}, \text{standard})$. The domain is finer. Is it continuous? We take a basic open set in the standard codomain, say $(a, b)$. Its preimage is just $(a,b)$. Is $(a,b)$ open in the lower-limit topology? Yes, because we can write it as a union of the basis elements of that topology. So, $f$ is continuous. A finer domain provides all the necessary open sets to be preimages.

• ​​Case 2:​​ $g: (\mathbb{R}, \text{standard}) \to (\mathbb{R}, \text{lower-limit})$. The domain is coarser. Is it continuous? We take a basic open set in the lower-limit codomain, say $[a, b)$. Its preimage is just $[a, b)$. Is this set open in the standard topology? No! No open interval can be squeezed around the point $a$ that stays within $[a, b)$. Because the domain's topology is too coarse—it lacks the necessary open set—the function $g$ is not continuous.

This principle resolves many puzzles. Continuity is not an absolute property of a formula like "$f(x)=x$". It's a statement about the relationship between two topological structures.

While the definition of continuity is global (it must hold for all open sets), the question of whether a function is continuous can be asked at a single point. This brings us back to some of the famous and beautifully pathological functions from real analysis.

Consider the ​​Dirichlet function​​, which is $1$ for rational numbers and $0$ for irrational numbers. Is this function continuous anywhere? Let's pick any point $x_0$. If $x_0$ is rational, $f(x_0)=1$. Let's choose a small open bubble around $1$ in the codomain that doesn't contain $0$, say the interval $(0.5, 1.5)$. For $f$ to be continuous at $x_0$, we'd need to find an open neighborhood around $x_0$ in the domain whose image lies entirely inside $(0.5, 1.5)$. But this is impossible! Any neighborhood around any rational number, no matter how tiny, will also contain irrational numbers. The function will map these points to $0$, which is outside our target bubble. The same logic applies if we start with an irrational point. The function's values jump so erratically that it's discontinuous at every single point.

A more subtle case is the function $f(x)$ which is $x$ for rational numbers and $-x$ for irrational numbers. Let's test it at a point $x_0 \neq 0$. The function values in any tiny neighborhood of $x_0$ will be close to both $x_0$ and $-x_0$. Since these are distinct numbers, the image of the neighborhood is "torn" into two pieces, and the function is not continuous. But something miraculous happens at $x_0=0$. As $x$ approaches $0$, both $x$ and $-x$ approach $0$. The potential "tear" in the function seals itself at this one point. You can indeed show that for any open bubble around $f(0)=0$, you can find a neighborhood around $0$ in the domain that maps entirely inside it. This function performs a magical feat: it is continuous at exactly one point, $x=0$, and discontinuous everywhere else.

Finally, continuous functions play nicely with common mathematical operations.

If a function $f: X \to Y$ is continuous, and you consider its ​​restriction​​ to a smaller subspace $A \subset X$, is the new function $f|_A: A \to Y$ still continuous? Yes, and the reason flows directly from the definitions. An open set in the subspace $A$ is just the intersection of $A$ with an open set from the larger space $X$. When we pull an open set $V$ from $Y$ back with $f|_A$, the result is precisely the intersection of $A$ with the preimage $f^{-1}(V)$. Since $f$ is continuous, $f^{-1}(V)$ is open in $X$. Its intersection with $A$ is therefore, by definition, open in $A$. Continuity is beautifully inherited by subspaces.

What about chaining functions? If $f: X \to Y$ and $g: Y \to Z$ are both continuous, what about the ​​composition​​ $g \circ f$? To check, we pull an open set $U$ from $Z$ all the way back to $X$. The first step, pulling it back through $g$, gives us the preimage $g^{-1}(U)$. Since $g$ is continuous, this is an open set in $Y$. The second step is to pull this new open set back through $f$. This gives us $f^{-1}(g^{-1}(U))$. Since $f$ is continuous, and $g^{-1}(U)$ is an open set in its domain, this final preimage is open in $X$. The composition of continuous functions is continuous.

But be careful! Does the reverse hold? If $g \circ f$ is continuous, must $f$ and $g$ be continuous? Not necessarily. A non-continuous function can have its "flaw" masked by a subsequent function. Imagine a function $f$ that creates a non-open preimage. If the next function $g$ is a constant function, the total composition $g \circ f$ will also be a constant function, and as we saw, constant functions are always continuous. The continuity of the composition tells you about the total journey, not necessarily about each individual step.

In moving from the $\epsilon-\delta$ world to the topological one, we have traded the ruler for a more fundamental tool: the very structure of the space itself. This abstract and powerful definition of continuity allows us to understand the deep connections between different mathematical objects in a way that is both elegant and astonishingly general. It is a testament to the beauty and unity of modern mathematics.

We have spent some time developing a rather abstract idea of continuity, one based on "open sets." You might be wondering, "Why go through all this trouble? What was wrong with the good old $\epsilon$-$\delta$ definition from calculus?" The answer, and I hope you will come to see the beauty in it, is that by stepping away from numbers and distances, we have created a tool of incredible power and generality. This new perspective allows us to see deep connections between seemingly unrelated parts of the mathematical world. It’s like having a new pair of glasses that reveals the hidden structural skeleton of reality. Let's put on these glasses and take a look around.

One of the most profound roles of continuity is to tell us what is not possible. A continuous function cannot arbitrarily tear a space apart. Imagine you have a single, connected object—say, a rubber band, which we can model as the interval $[0, 1]$. Now, suppose you want to map every point on this rubber band to one of two separate locations, let's call them "A" and "B". Can you do this continuously?

Our topological intuition screams "No!". To do so, you'd have to "break" the rubber band somewhere, sending one piece to A and the other to B. The point of breakage would be a discontinuity. Our abstract definition confirms this intuition perfectly. If our target space consists of a set of isolated, discrete points (like the integers $\mathbb{Z}$ with the discrete topology), then any continuous function from a connected space (like our interval) must be a constant function. It must map the entire rubber band to a single point, say A. It cannot split it up. This is because the image of a connected space under a continuous map must also be connected. Since the only connected subsets of a discrete space are single points, the image must be a single point!

This idea generalizes beautifully. If you have a continuous function mapping any space $X$ to a discrete space $Y$, the function is continuous if and only if it's "locally constant." This means that for every point in the domain, there's a small open neighborhood around it where the function doesn't change its value. In essence, the function must partition the domain space $X$ into a collection of disjoint open sets, with each open set being mapped to a single point in $Y$. The topology of the target space places a powerful constraint on the types of continuous behaviors we can observe.

Beyond telling us what we can't do, continuity is the master blueprint for building new mathematical worlds from old ones. When we construct a more complex space, we almost always do so by ensuring that the natural maps related to the construction are continuous.

Think about the Cartesian plane, $\mathbb{R}^2$. We think of it as a product of two real lines, $\mathbb{R} \times \mathbb{R}$. How do we define "nearness" in the plane? The most natural way is to ensure that the simple act of looking at a point's coordinates is a continuous process. The function that takes a point $(x,y)$ and gives you back its $x$-coordinate should be continuous, and the same for the $y$-coordinate. These functions are called projections. The product topology is defined as precisely the "most economical" topology—the one with the fewest possible open sets—that guarantees all these projection maps are continuous. This isn't just a trick for $\mathbb{R}^2$; it's the fundamental principle used to define topologies on products of any spaces, including the infinite-dimensional spaces that form the bedrock of functional analysis.

What if we want to build not by multiplying, but by gluing? Imagine taking a rectangular strip of paper and gluing the ends together. If you glue them straight, you get a cylinder. If you put a half-twist in before gluing, you get a Möbius strip. In topology, this "gluing" is formalized by defining an equivalence relation on the original space (the rectangle) and considering the set of equivalence classes. The resulting space is called a quotient space. But what is its topology? Again, we let continuity be our guide. The quotient topology is defined as the "richest" topology—the one with the most possible open sets—that ensures the natural projection map sending each point to its "glued" equivalence class is continuous. This single principle is the engine of creation in algebraic topology, allowing us to construct spheres, tori, and all manner of exotic shapes by intelligently gluing together simpler pieces.

The true power of the topological view of continuity is its role as a unifying concept, a common language spoken by many different fields of mathematics.

​​Real and Complex Analysis:​​ In your first calculus course, you learned that continuous functions are "nice." You might have assumed this niceness means they carry open intervals to open intervals. But this is a subtle trap! Consider the simple continuous function $f(x) = (2x-1)^2$ on the real line. If we look at the image of the open interval $(0, 1)$, we find that it is the interval $[0, 1)$. This set is not open because it includes its left endpoint, $0$. The image of an open set under a continuous function is not always open!. A function that does have this special property is called an ​​open map​​.

Now, let's step into the world of complex numbers. Here, something magical happens. A function that is differentiable in the complex sense (which we call "analytic") is incredibly rigid and powerful. The ​​Open Mapping Theorem​​, a cornerstone of complex analysis, tells us that any non-constant analytic function is an open map! This is a profound difference from real functions. One beautiful consequence is that if an analytic function is one-to-one, its inverse function is guaranteed to be continuous. Why? Because to check the continuity of the inverse, we need to see if the preimage of an open set is open. But the preimage under the inverse is just the image under the original function, which we now know is open thanks to the theorem!

​​Algebra and Group Theory:​​ Let's swing over to the abstract world of group theory. A ​​topological group​​ is a group that is also a topological space, where the group operations (multiplication and inversion) are continuous. This marriage of algebra and topology gives rise to a rich theory. Now, what if we take any group and put the discrete topology on it, where every point is its own little open set? A funny thing happens: every function from a discrete space is continuous! This means the multiplication and inversion maps are automatically continuous, and our group becomes a topological group for free, no matter what its algebraic structure is. This "discrete case" serves as a fundamental example and a baseline for comparison when studying more intricate topological groups like matrix groups.

​​Measure Theory:​​ This interplay with discrete spaces has interesting consequences elsewhere. In measure theory, ​​Lusin's Theorem​​ is a deep result stating that any measurable function on a nice space can be approximated by a continuous function. That is, you can find a continuous function that agrees with your original function everywhere except on a set of arbitrarily small measure. But what happens if our space has the discrete topology? Well, as we've seen, any function on this space is already continuous. So, if we are asked to find a continuous approximation $g$ for a function $f$, we can simply choose $g=f$! The set where they differ is empty, and its measure is zero. The powerful Lusin's Theorem becomes trivially true, its challenge completely defused by the nature of the underlying topology.

​​Differential Geometry:​​ On a smooth manifold—the mathematical object used to model everything from the surface of a soap bubble to the spacetime of general relativity—we have two natural ways to think about "open sets." First, there's the topology that comes from its construction via charts and atlases, the "atlas topology." Second, if the manifold has a Riemannian metric (a way to measure distances), we can define a topology using open balls, just like in a standard metric space. For the entire edifice of differential geometry to work, these two topologies must be the same. The formal way to say this is that the identity map from the atlas topology to the metric topology, and its inverse, must both be continuous. This ensures that our local notion of "nearness" from the charts aligns perfectly with our global notion of "nearness" from the distance function, allowing calculus and geometry to coexist peacefully.

To close, let me show you an example that demonstrates just how counter-intuitive and powerful continuity can be. Consider the Cantor set, that famous fractal constructed by repeatedly removing the middle third of intervals. It's a "dust" of points; it contains no intervals and seems almost one-dimensional, if not zero-dimensional. Now consider a solid two-dimensional square.

Is it possible to find a continuous surjective function that maps the Cantor set onto the entire square? The answer, astonishingly, is yes. There exist "space-filling curves" that do just this. A one-dimensional, totally disconnected dust of points can be continuously squashed and smeared to cover every single point of a 2D square. This shatters our simple intuitions about dimension. If we then ask, what does the preimage of a point inside the square look like? That is, which collection of points in the Cantor set all get mapped to the same point in the square? For any such map, this set of points must be a non-empty, compact, and totally disconnected set—a little "Cantor-like" set in its own right.

This is the world that the topological definition of continuity opens up for us. It is a world of surprising connections, deep structural truths, and mind-bending possibilities. It is the language that allows us to describe the shape of space itself.