# Motivation behind topology

What is the motivation behind topology?

For instance, in real analysis, we are interested in rigorously studying about limits so that we can use them appropriately. Similarly, in number theory, we are interested in patterns and structure possessed by algebraic integers and algebraic prime numbers.

Some googling and wiki-ing gave me that topology studies about deformation of objects in some space i.e. how an object in some space behaves under a continuous map. However, when I started reading the subject it starts of by defining what a topology is i.e. a set of subsets of a set with certain properties. I fail to see the connection immediately. I would appreciate if someone could give a short bird's eye view of topology.

-
Here's a relevant question on MathOverflow. –  Dylan Moreland Aug 27 '11 at 18:33
First study metric spaces as a generalization of real analysis. Once you understand what concepts like limits, compactness, and connectedness mean in metric spaces and the pervasive role of open balls in metric spaces you will see the motivation for definitions in general topology in terms of "open sets". –  KCd Aug 28 '11 at 3:09
@KCd I think he is asking about "motivation behind topology" in a sense in which "what made people study this subject?".In a historical perspective the point set topology we are studying emerged much later(some 20-30yrs) than algebraic topology. The basic motive to consider topological spaces for Poincare seems to be something related to differential equations. Even Riemann's monodromy theorem have a nice connection with fundamental group. However the notion of topological space and its off springs namely "fundamental group" proved very fundamental as time went on. –  Dinesh Sep 4 '11 at 22:03
cnt..The metric space study before landing into topology proper may be much like a pedagogical motive to get comfortable thought process rather than the "real motive" behind doing topology. –  Dinesh Sep 4 '11 at 22:06

Topology can mean different things in mathematics, depending on the context.

If a mathematician describes themselves as a topologist, this likely means that they study various kinds of shapes (technically, manifolds, or related kinds of spaces), with an eye perhaps to classifying them (a typical example being the classification of closed orientable surfaces via the number of handles attached to a sphere), or understanding certain invariants or other aspects of their structure (e.g. the Poincare conjecture, which gives a characterization of the three-dimensional sphere in terms of a certain invariant, namely its fundamental group).

On the other hand, the basic axioms of topology that you are asking about (a topology on a space is a collection of subsets, called open subsets, satisfying the following axioms ...) are much more general. There is a branch of mathematics that focuses on studying (more or less) these axioms and their consequences (appropriately known as general topology, where general refers to the generality of the axioms), but this investigation is rather different in flavour to the mathematics described in the preceding paragraph --- it is perhaps closer to set theory than it is to the investigation of the topology of manifolds and so on.

But thinking in terms of the subject of general topology is not a good way to understand the point of the axioms of topology. By analogy, consider the group axioms: there is a well-developed area of mathematics called group theory which studies (more or less) these axioms and their consequences, but the notion of group is ubiquitous in mathematics, and plays a fundamental role in many areas of mathematics besides group theory proper (including number theory, topology, and geometry). Thus the group axioms capture a concept which is of fundamental mathematical importance, and this is why we isolate the notion of group and study it.

Similary, the axioms of a topology, and the basic definitions associated to them (such as continuity of maps, and connectedness and compactness of subsets) were formulated as the consequence of a rather long attempt during the 19th and early 20th centuries to isolate and abstract the basic notions related to continuity and limits. These concepts are again ubiquitous in mathematics, and so the notion of topology (in the sense of the axioms you asked about) are applied in an enormous number of different contexts throughout mathematics (not just in those particular areas studied by topologists or general topologists, but in geometry, analysis, number theory, parts of algebra, logic, ...). Although the axioms may seem unusual, and unrelated to the notions of shape and position that you see discussed in an intuitive account of topology, they are in fact carefully crafted to capture, in very general language, the notions of continuity, nearness, limits, and so on.

In order to see the truth of my last claim, you will have to invest some time studying the axioms (i.e. learning some basic general topology), and then see how it is applied in various contexts. At least in the U.S., this typically happens in advanced undergraduate and beginning graduate courses.

-
I think that the last two paragraphs are about as honest a short answer as is possible. –  Brian M. Scott Aug 28 '11 at 2:15
Thanks. I am learning topology from the book Topology without tears. So I am spending some time on the basics by working out the exercise problems. On a similar note, I wanted to ask another question, what is the reason for defining a topology as one which is closed under arbitrary unions and finite intersection. I can see the motivation comes from open sets on $\mathbb{R}^n$ but why are arbitrary unions and finite intersections the defining property? Should I ask this as a different question? –  Adhvaitha Aug 28 '11 at 3:57
I think the question I have mentioned in my previous comment has already been asked here. (math.stackexchange.com/questions/4114/…) –  Adhvaitha Aug 28 '11 at 4:05
@Adhvaitha: The short answer is that that’s what’s needed to make the definition useful. However, you might take a look at the MathOverflow question that Dylan Moreland mentioned. Some of the answers, especially this one by Marcos Cossarini, specifically offer intuitions about the asymmetry between unions and intersections. –  Brian M. Scott Aug 28 '11 at 4:06

This math.SE question may be relevant, but not pedagogically optimal.

Pedagogically I think the simplest answer is to axiomatize topological spaces via the Kuratowski closure axioms. Instead of specifying what properties open sets or closed sets satisfy, the Kuratowski closure axioms specify a closure operator $S \mapsto \text{cl}(S)$ on subsets $S$ of a set $X$ and axioms it ought to satisfy. You should think of closure as axiomatizing abstract limits (that is, $\text{cl}(S)$ roughly corresponds to the set of all possible limits of sequences of elements of $S$, but it actually doesn't; we need to replace sequences by filters or nets in full generality). Then the closed sets are precisely the ones for which $S = \text{cl}(S)$ and the open sets are the complements of the closed sets as usual.

It's worth mentioning that the notion of a topological space is absurdly general. For many applications you'll only need to think about the topology of much more restricted types of spaces (e.g. manifolds, CW-complexes...). Nevertheless, because it is so general, it can be fruitfully applied to many areas of mathematics, and because the set of axioms used is relatively sparse, proofs in full generality are generally relatively short.

-

This will answer only the question in the second paragraph.

Imagine the half-open interval $[0,1)$ with the usual open sets. In particular, an open neighborhood of $0$ contains $0$ itself and all positive numbers sufficiently close to $0$.

Now alter the definition of "open set", so that every open neighborhood of $0$ contains not only $0$ and all positive numbers close enough to $0$, but also all numbers close enough to $1$. I.e. $[0,\varepsilon)\cup(1-\eta,1)$.

Can you see how that alters the way in which the whole space is connected together?

Thus: how the space is connected together, is simply a matter of which sets are open. That's the connection between the two things.

-
Sorry. I don't see what you are coming at. In the first case, the neighborhood of $0$ are sets of the form $[0,\epsilon)$ and in the second case the neighborhoods are $[0,\epsilon) \cup (1-\eta,1)$. So in the first case all the neighborhoods are connected whereas in the second one they are not? Does this mean the space described in the first case is connected while in the second case the space is disconnected? Thanks. –  Adhvaitha Aug 28 '11 at 4:41

This may not be quite the answer you are looking for but I think a "birds-eye" view of topology is somewhat hard. One example which I found somewhat enlightening though, was when I first thought about the topology of Baire space.

The topology is defined on the set of infinite sequences over the natural numbers. The topology is then defined to be such that each open set is given by the set of all sequences which coincide on some finite set of points. This can be made somewhat simpler by saying that the topology is generated by a basis of open sets given by finite sequences, to each finite sequence you associate the set of sequences beginning with this finite sequence, and then let the topology be given by all unions and intersections of such sets.

A continuous function from this Baire space into the the real numbers (with the metric topology, that is to say the topology generated by open intervals) now correspond, intuitively speaking, to those functions such that given "enough" information about he sequence (that is to say, a large enough initial segment) you get a "good enough" approximation of where the value of the whole sequence will be (i.e. a small interval within which the value will be).

Of course you can choose the codomain to be, again, Baire space. In which case the continuous functions correspond to those where an arbitrary large initial segment of the "output" can be generated by giving a large enough initial segment of the "input". Such continuous functions are, thus, in a sense, the computable functions, since they do not "get stuck".

I rather liked this example since it indicates a way in which topology gives a sense of "information content" and things behaving well with respect to approximation. I guess this language is more close to topology as its studied in areas as domain theory, but I figured it might help to consider it from a somewhat non-geometric perspective. Also it suggest, as so often seems to be the case with mathematics, that the interesting things happen when you start consider structure preserving maps, and that the definition of a topology is, perhaps, more interesting as a way to be able to describe the notion of continuity of a function.

(As a remark that I am aware that the Baire topology is a metric topology, but I feel that the definition in terms of a topology is much more natural than in terms of the ultrametric distance)

-