Saturated Open Set

In the mathematical field of topology, one of the most powerful techniques for creating new spaces is to take an existing one and "glue" parts of it together. This intuitive process, which can turn a flat square into a cylinder or a donut, gives rise to what are known as quotient spaces. However, this act of creation poses a fundamental challenge: how do we define the topological structure, such as the notion of an open set, on this newly formed object? How do we ensure that the concept of "nearness" is coherently transferred from the original space to the new one?

This article addresses this question by exploring the central role of the ​​saturated open set​​. This single concept provides the complete blueprint for understanding the topology of any quotient space. In the following chapters, we will first delve into the ​​Principles and Mechanisms​​, defining what a saturated set is and how it governs the structure of quotient spaces through clear examples. Subsequently, in ​​Applications and Interdisciplinary Connections​​, we will see how this theoretical tool is used to construct a diverse menagerie of spaces, from well-behaved manifolds to strange pathological curiosities, revealing the profound power and subtlety of topological gluing.

Imagine you have a flat strip of paper. You can bend it and glue the two ends together to make a circle. Or you can take a square sheet, glue one pair of opposite edges to make a cylinder, and then, if you were working in four dimensions, you could bend the cylinder around and glue its two circular ends to make a torus—the surface of a donut. This intuitive act of "gluing" is one of the most powerful ideas in mathematics for creating new shapes from old ones. In topology, we call these new shapes ​​quotient spaces​​.

But this raises a profound question. The original space, like the flat sheet of paper, had a well-defined notion of "nearness". Any point had a small open disk around it. After we glue things together, what does it mean for points to be "near" each other in the new shape? How do we define an "open set" on the surface of our newly-minted torus? The answer is both elegant and surprisingly simple, and it all revolves around a concept called a ​​saturated set​​.

Let's call our original space $X$ (the paper sheet) and our new, glued-up space $Y$ (the torus). The "gluing instructions" are given by an equivalence relation, and the map that takes a point in $X$ to the point in $Y$ it becomes is called the ​​quotient map​​, let's call it $p$. The fundamental rule for defining the topology on $Y$ is this:

A subset $V$ in the new space $Y$ is declared ​​open​​ if, and only if, its ​​preimage​​, $p^{-1}(V)$, is an open set in the original space $X$.

At first glance, this might seem like we're just passing the buck. But this single definition is the key that unlocks everything. It forces us to ask: what kind of sets in our original space $X$ can possibly be the preimage of some set in $Y$?

Think about the gluing. When we identify two points $x_1$ and $x_2$ in $X$, they become the same point $y = p(x_1) = p(x_2)$ in $Y$. Now, if we take any set $V$ in $Y$ that contains this point $y$, its preimage $p^{-1}(V)$ in $X$ must contain both $x_1$ and $x_2$. It has no choice! This leads us to a crucial property. A set in the original space that is a full preimage of some set in the new space is called a ​​saturated set​​.

A subset $A$ of our original space $X$ is ​​saturated​​ if, for every point $x$ in $A$, the entire equivalence class of $x$ (all the other points that get glued to $x$) is also contained in $A$. A saturated set completely respects the gluing instructions. If it grabs one member of a "glued family," it must grab them all.

Let's make this concrete. Imagine the 2-sphere $S^2$ (the surface of a ball). We can create a bizarre but important space called the real projective plane, $\mathbb{R}P^2$, by gluing every point $\vec{v}$ to its antipode, $-\vec{v}$. The equivalence classes are pairs of opposite points $\{\vec{v}, -\vec{v}\}$. Now, consider the open northern hemisphere, $H = \{ (x, y, z) \in S^2 \mid z > 0 \}$. Is this set saturated? Let's check. Take any point $\vec{v}$ in $H$. Its antipode, $-\vec{v}$, has a negative $z$-coordinate, so it lies in the southern hemisphere. Since $-\vec{v}$ is not in $H$, the set $H$ fails the saturation test. It doesn't respect the gluing rule.

For a set on the sphere to be saturated under this antipodal identification, it must be perfectly symmetric. If it contains a point, it must also contain its opposite. An open "belt" around the equator, for example, would be a saturated set.

Now we can combine our two ideas. The open sets in the new space $Y$ are the images of sets in the original space $X$ that are both ​​open​​ and ​​saturated​​. This is the central mechanism of the quotient topology. To understand what's open in the glued space, we just need to find the open sets in the original space that respect the gluing instructions.

Let's put on our hard hats and see how this principle allows us to build and understand new spaces.

​​The Circle from an Interval:​​ Let's go back to our simplest example: making a circle $S^1$ by gluing the endpoints of the interval $X = [0,1]$. The equivalence relation identifies $0$ and $1$. The only non-trivial equivalence class is $\{0, 1\}$. What do the saturated open sets in $[0,1]$ look like?

• Any open interval $(a,b)$ where $0 a b 1$ doesn't contain $0$ or $1$. So it's trivially saturated. Its image on the circle is a simple open arc.
• But what about a neighborhood of the point where $0$ and $1$ are glued together? A set like $[0, c)$ for some small $c>0$ is open in the topology of $[0,1]$, but it's not saturated because it contains $0$ but not $1$. So its image is not an open set in the circle. To make a saturated open set containing the endpoints, we must take a piece from both ends. A set of the form $[0, c) \cup (d, 1]$ is both open in $[0,1]$ and saturated, because it contains both $0$ and $1$. Its image under the gluing map is a proper open arc on the circle that contains the "seam". This collection of open arcs inside the circle and open arcs that cross the seam forms a complete basis for the circle's topology.

​​The Cylinder and the Torus:​​ Now let's build a cylinder by taking the plane $\mathbb{R}^2$ and identifying any two points $(x_1, y_1)$ and $(x_2, y_2)$ if $y_1=y_2$ and their x-coordinates differ by an integer. This is like rolling the plane into an infinitely long tube. A set $A \subset \mathbb{R}^2$ is saturated if whenever $(x,y) \in A$, then $(x+n, y) \in A$ for all integers $n$. The set must be infinitely repeating along the x-axis. Consider the open rectangle $A_1 = (0, 1) \times (-1, 1)$. It's not saturated. Its saturation is the union of all its integer translates: $\bigcup_{n \in \mathbb{Z}} (n, n+1) \times (-1,1)$. This resulting set is a collection of disjoint open strips, so it's open in $\mathbb{R}^2$. Therefore, the image of the original rectangle $A_1$ is an open set on the cylinder. Fascinatingly, even a half-open set like $A_2 = [0, 1) \times (-1, 1)$ can have an open image. Its saturation is $\bigcup_{n \in \mathbb{Z}} [n, n+1) \times (-1,1)$, which perfectly tiles the entire infinite strip $\mathbb{R} \times (-1,1)$. Since this saturation is an open set, the image of $A_2$ is also open in the cylinder! The same "wraparound" logic applies when building a torus, but you have to satisfy the saturation condition in both the horizontal and vertical directions simultaneously.

The power of quotient spaces isn't limited to gluing edges. We can perform much more radical surgeries.

​​Collapsing to a Point:​​ Imagine taking the entire real number line $\mathbb{R}$ and collapsing the infinite, discrete set of integers $\mathbb{Z}$ into a single point. All the integers get glued together, while every other number remains distinct. What does an open neighborhood of this new "super-point" look like in the quotient space? Its preimage must be a saturated open set in $\mathbb{R}$, which means it must be an open set that contains all the integers. A simple example would be a union of tiny open intervals centered on each integer: $\bigcup_{n \in \mathbb{Z}} (n - \epsilon, n + \epsilon)$. This reveals something amazing: in the new space, points that were arbitrarily far apart in $\mathbb{R}$ (like $1$ and $1,000,000$) have now become part of the same entity, and any open set containing that entity must also contain points that were originally close to any of the integers. This is why this "super-point" lies in the boundary of the rest of the space.

​​Projecting onto a Circle:​​ Let's take the plane with the origin removed, $X = \mathbb{R}^2 \setminus \{(0,0)\}$, and say two points are equivalent if they lie on the same open ray from the origin. This essentially collapses each ray down to a single point. The set of these rays can be identified with the unit circle, $S^1$. Is the unit circle $S^1$ itself a saturated set in $X$? No. For any point on the circle, its equivalence class is the entire ray passing through it. The circle only contains one point from that ray, not the whole thing. A saturated set would have to be a cone-like region. The circle $S^1$ is what we call a ​​section​​ or ​​transversal​​: a set that picks out exactly one representative from each equivalence class. This is a very different, but equally important, concept from a saturated set.

From making circles and donuts to collapsing infinite sets of points, all these seemingly different constructions are governed by one unified principle. The structure of the new space is completely determined by the ​​saturated open sets​​ of the original space. This idea gives us a precise and powerful lens through which we can understand the topology of shape. It tells us that the properties of our glued-up world are inherited directly from the properties of the original world, but only via those subsets that fully respect the blueprint of our gluing. Even more, this principle is local: if we look at just one saturated open piece of our original space, its mapping to the corresponding piece of the new space behaves just like a full quotient map in its own right. It is a beautiful testament to how a single, well-chosen abstract definition can bring clarity and order to a vast universe of geometric forms.

We have spent some time developing the machinery of quotient spaces, with the concept of the "saturated open set" as our trusty guide. You might be wondering, "What is all this for?" Is it just an abstract game of definitions? Far from it. This machinery is, in fact, a powerful architect's toolkit. It allows us to take existing spaces, which we can think of as our raw materials, and "glue" them together along prescribed seams to construct entirely new universes.

The saturated open sets are the blueprints for this construction. They dictate the properties of the new world we’ve built. Sometimes, the result is familiar and well-behaved. Other times, the worlds we create are fantastically strange, revealing deep truths about the nature of space itself. Let's embark on a journey through this gallery of creations, to see the power and the peril of topological gluing.

Imagine you have a flexible, flat sheet of paper, a perfect model of a piece of the Euclidean plane $[0,1] \times [0,1]$. It’s a nice, respectable space—it's Hausdorff, meaning any two distinct points can be put in their own little "neighborhood bubbles" that don't touch. What happens if we decide to glue the left edge of the square to the right edge? We are defining an equivalence relation: a point $(0, y)$ on the left edge is now considered the same as the point $(1, y)$ on the right edge.

The result, as you can guess, is a cylinder. But is this new, curved space as "nice" as the flat square we started with? Is it still Hausdorff? This is not a trivial question! The points along the seam where we glued are special. How do we know we can still separate two distinct points on this seam?

This is where our blueprints, the saturated open sets, come to the rescue. To separate two points $z_1$ and $z_2$ on the seam, we need to find disjoint open neighborhoods for them in the cylinder. This is equivalent to finding two disjoint saturated open sets back in the original square that contain the preimages of $z_1$ and $z_2$. A set in the square is saturated if, whenever it contains a point $(0,y)$, it must also contain its identified partner $(1,y)$.

Aha! We can simply take thin, horizontal "bands" across the square. If we build a band around the height of $z_1$ and another around the height of $z_2$, and make them thin enough, they won't overlap. Each band is automatically saturated because it stretches all the way from the left edge to the right edge. The images of these disjoint saturated bands in the quotient space become the disjoint neighborhood bubbles we were looking for on the cylinder. With a bit of care, we can prove that this works for any pair of distinct points. So, our cylinder is indeed Hausdorff. The art of gluing has successfully produced a familiar, well-behaved object. The same idea allows us to build tori (doughnuts), Möbius strips, and other fundamental objects in geometry.

The success of building a cylinder might give us a false sense of security. The rules of gluing are subtle, and slight changes can lead to dramatically different, even pathological, outcomes.

The Birth of Inseparable Twins: The Line with Two Origins

Let's try a different construction. We take two separate, parallel copies of the real line, $\mathbb{R} \times \{1\}$ and $\mathbb{R} \times \{2\}$. Now, let's glue them together everywhere except at the origin. That is, for any non-zero number $x$, we declare the point $(x,1)$ on the first line to be the same as the point $(x,2)$ on the second. But we leave the two origins, $(0,1)$ and $(0,2)$, as distinct points. The resulting space is famously known as the "line with two origins".

Let's call our two origins $p_1$ and $p_2$. Are they separable in the Hausdorff sense? Can we put them in their own non-overlapping bubbles? Let's try to build a neighborhood bubble $U_1$ around $p_1$. Its preimage in our starting space must be a saturated open set containing $(0,1)$. Since it's an open set around $(0,1)$, it must contain a small interval $(-\epsilon, \epsilon)$ on the first line. But for the set to be saturated, for every tiny non-zero number $x$ in this interval, it must also contain the point $(x,2)$ on the second line. So, any neighborhood bubble around $p_1$, no matter how small, inevitably includes points that are "glued" to points right next to $p_2$.

The same thing happens if we try to build a bubble $U_2$ around $p_2$. It will inevitably snatch up points glued to points right next to $p_1$. The conclusion is inescapable: any open set containing $p_1$ must have a non-empty intersection with any open set containing $p_2$. Our two origins are topologically inseparable.

This space fails to be Hausdorff, and this single failure has profound consequences. In differential geometry, a manifold is a space that locally looks like Euclidean space $\mathbb{R}^n$. Our line with two origins actually satisfies this local property everywhere! Yet, because it is not Hausdorff, it is denied entry into the club of manifolds. This single example teaches us that the local structure is not enough; a coherent global structure is essential. It also shows that this can happen even when the sets we are identifying are perfectly fine closed sets in the original space.

The Infinite Bouquet: Losing Compactness

Let's return to a single real line, $\mathbb{R}$. This time, our equivalence relation is more dramatic: we declare all integers $\mathbb{Z} = \{\dots, -2, -1, 0, 1, 2, \dots\}$ to be a single point, while leaving all other points distinct. What does this space look like? You can imagine taking the real line and, for each interval $(n, n+1)$, tying the ends $n$ and $n+1$ together at the single point where all integers now live. The result is an infinite "bouquet" of loops, all joined at one central point.

This space, it turns out, is still Hausdorff! We can separate any two non-integer points easily. We can even separate a non-integer point from the central "integer point." But has it inherited all of $\mathbb{R}$'s properties? The real line is not compact. What about our bouquet? It is also not compact. We can see this intuitively: consider an open cover consisting of one small open set containing the central point, and an infinite collection of open sets, each covering one of the loops. You can't throw away any of the sets covering the loops, so no finite subcover exists. This example shows that different gluing rules have different consequences—we can lose one property (like compactness) while preserving another (like the Hausdorff property).

The Rational Singularity: When Density Creates Chaos

Now for our most mind-bending creation. Again, we start with $\mathbb{R}$. This time, we identify all rational numbers $\mathbb{Q}$ to a single point. Recall that the rationals are dense in the real line; between any two irrationals, there is a rational. What kind of universe does this violent act of gluing create?

Let's call the point representing all of $\mathbb{Q}$ the "rational singularity," $p_{\mathbb{Q}}$, and let's take any other point, corresponding to an irrational number $y$. Can we separate them? Let's try to build a neighborhood bubble around $y$. This bubble must contain an open interval around $y$. But because the rationals are dense, any open interval, no matter how small, is guaranteed to contain a rational number. This means any open neighborhood of $y$ must contain a point that gets glued into $p_{\mathbb{Q}}$. Therefore, every open set containing the point for $y$ also contains the point $p_{\mathbb{Q}}$! The space is catastrophically non-Hausdorff. The irrational points are all "stuck" to the rational singularity.

Yet, amidst this chaos, one property miraculously survives: connectedness. The resulting space is still connected. Why? Because if you could split the quotient space into two disjoint open pieces, their preimages would form two disjoint, non-empty, saturated open sets that cover all of $\mathbb{R}$. But a set is saturated under this relation only if it contains all of $\mathbb{Q}$ or none of it. If we assume the preimage of one piece contains $\mathbb{Q}$, then its complement (the preimage of the other piece) must consist only of irrationals. A non-empty open set made entirely of irrationals is impossible in $\mathbb{R}$! So, $\mathbb{R}$ cannot be split this way, which means our quotient space cannot be split either. The original connectedness of the real line is robust enough to survive even this radical surgery.

The concept of a quotient space is not just a topologist's curiosity; it is a fundamental language for describing symmetry in physics and mathematics. Consider a group of transformations, or symmetries, acting on a space. For example, the group of rotations acting on a disk. The set of all points that can be reached from a single point $x$ by applying these transformations is called the orbit of $x$. We can then ask: what does the space of orbits look like? This is, by its very definition, a quotient space, where the equivalence relation is "being in the same orbit."

Let's take the group of non-zero real numbers $\mathbb{R}^*$ acting on the plane $\mathbb{R}^2$ by scalar multiplication. This action scales vectors, stretching or shrinking them. The orbit of any non-zero vector $\mathbf{v}$ is the full line passing through the origin (with the origin itself removed). The orbit of the origin is just the origin itself.

The orbit space, then, consists of points representing lines through the origin, plus one special point representing the origin's orbit. Let's call this special point $p_0$. What does a neighborhood of $p_0$ look like? A saturated open set containing the origin must contain a small open disk centered at the origin. But this disk contains points from every line through the origin! This means any open set in the orbit space that contains $p_0$ must also contain every other point. The point $p_0$ is in the closure of every other point. The space is not even $T_1$, a weaker condition than Hausdorff. This strange topological behavior arises naturally from a very simple geometric action.

From the simple cylinder to the bewildering rational singularity, the quotient construction allows us to build a vast and fascinating universe of topological spaces. The concept of the saturated open set is the master key that unlocks their properties. It is the blueprint that tells us whether our new creation will be a familiar manifold or a pathological curiosity.

This journey reveals the profound unity of mathematics. An abstract definition from general topology becomes a practical tool for geometric construction, a language for describing physical symmetries, and a lens for understanding the deep and often surprising consequences of identifying points. The beauty lies not just in the strange objects we can create, but in our ability to predict their properties and understand why they behave the way they do, all by carefully examining the structure of our blueprints.