7

Question: Is a topological space considered to be a class in set theory?

So, by the might of wikipedia, I've found that

...a topological space may be defined as a set of points, along with a set of neighbourhoods for each point...

This sounds like a set of sets, because for every point element there is a neighborhood set. Does this make it a class in set theory, as shown here:

In set theory and its applications throughout mathematics, a class is a collection of sets (or sometimes other mathematical objects) that can be unambiguously defined by a property that all its members share.

Then in order to understand what a class is, via ncatlab, I found what a collection is here:

use the word collection to denote a bunch of “things”

Which sounds self-referential. It seems like that the idea of an element and set are fairly concrete, and then a class/collection is just a bag that you throw things in.

On a slight tangent, how do you define higher level classes in set theory? Is there any notation that can show that this is a class of a class?

multicusp
  • 89
  • 5
    A set of sets isn't a problem; in set theory, every mathematical object other than the empty set is a set of sets. The paradox arises from considering (among other candidates) the set of all sets. – chepner Jul 28 '19 at 16:55

3 Answers3

23

A topological space is an ordered pair $(X, \mathcal T) $ where $X$ is a set and $\mathcal T$ is a set of subsets of $X$ satisfying certain axioms. An ordered pair of sets is itself a set. It is a class in the sense that every set is a class, but it has no special reason to be called a class instead of a set.

Matt Samuel
  • 58,164
20

Sets can contain sets. In the usual system of set theory, everything is a set - we start with the emptyset and "build up" (for a technical statement and proof of this, see e.g. here). For instance, we identify $0$ with $\emptyset$, $1$ with $\{\emptyset\}$, $2$ with $\{\emptyset,\{\emptyset\}\}$, and so on. Similarly, ordered pairs are defined purely set theoretically. So no, a topological space is still a set.

Classes enter the picture when we want to talk about collections - or if you prefer, properties - which cannot correspond to sets. For example, there cannot be a set of all sets not containing themselves - put another way, the collection of sets which don't contain themselves is a proper class. (Every set is a class but not conversely; we use the term "proper class" to refer to a class which is not a set.)

There are "higher set theories" which treat sets and classes - for instance, NBG. They have their own analogues of "non-set collections," generally called hyperclasses, and so on - Russell's paradox tells us that however far we go we'll always have reasonably-definable properties which don't correspond to objects in our framework.

Noah Schweber
  • 245,398
  • Part 1/4: I understand sets can contain sets. However, what I don't understand is the idea that there doesn't need to be a descriptor stating that a set contains another set, that are referred to as elements. It seems like the given method is obfuscating the constituent parts.

    Why not state the maximum depth of a set to the most fundamental element then? For example 2 being a 2-set, which would be defined as the set of a set of a fundamental element that can only be defined via definition.

     – multicusp Jul 27 '19 at 23:55
  • Part 2/4: A topological space could also be described as a set of a set, a collection of a set, a collection, a class of a collection, a class of a set, a class of a set of a set as well. The idea behind a collection is pointless. It allows for immense inclusivity at the cost of utility by having a purposely vague definition. – multicusp Jul 27 '19 at 23:55
  • Part 3/4: The idea that collections are, distinctly, properties seems silly. It frames the argument as though collections mean anything other than what you want. A collection can be a set, a collection can be a property, it can be anything. Fundamentally, it appears to mean nothing. – multicusp Jul 27 '19 at 23:56
  • Part 4/4: The "non-set collections" refer to a collection of classes, no? In which case the classes refer to a collection of sets. Set theory feels like it's "turtles all the way down". – multicusp Jul 27 '19 at 23:56
  • "Why not state the maximum depth of a set to the most fundamental element then?" That's essentially the rank. "It allows for immense inclusivity at the cost of utility by having a purposely vague definition." We need a term to use informally to help motivate the way we define formal terms. As to properties, I was actually getting at something rather technical - when we really get formal about this, we do talk about properties using model theory. (cont'd) – Noah Schweber Jul 28 '19 at 00:20
  • Beyond this I don't really know what to say. I'm sorry you dislike the way we tend to phrase things informally in set theory; it might be best to study it entirely formally at least for now, so you can convince yourself that it's not turtles all the way down. But I don't know exactly what you want from us at this point. – Noah Schweber Jul 28 '19 at 00:21
  • Besides, this all is completely tangential: the answer to your question "Is a topological space considered to be a class in set theory?" is "No (except in the sense that all sets are classes) - they are sets." And this much should be clear. – Noah Schweber Jul 28 '19 at 00:31
  • Let us continue this discussion in chat. – multicusp Jul 28 '19 at 00:31
4

I believe your question has been adequately answered, but I will iterate: A topological space is a pair consisting of a set of points and a subset of the powerset of the set of points with the property that each element of the set of points is contained in at least one element of this subset. (I.e., each point is present in at least one subset of the set of points that is deemed "open".) So the two members of the pair of a topological space are both classes, but only in the boring sense: all sets are classes.

It is worth pointing out that the set of "points" need not be points in what is normally considered a space. There is a surprising topological proof of the infinitude of primes where the topological space is the set of integers and the open sets are congruence classes of integers modulo various integers. One difficulty with your idea of specifying the "depth" of a set is that the integers, in their usual set theoretic model, has infinite depth, since incrementing an integer increases the depth by one. So as soon as you have the integers, you have infinite depth and then when you want to talk about actually large sets, annotating set depth becomes useless visual clutter.

In comments to another post, you object to defining classes via properties. Perhaps it would help if you understood how that is different from a thing being a set. A thing is a set if is constructed according to the axioms of your set theory. In, for instance ZF (abbreviating mightily), you get the empty set, an infinite set, and a few (fairly familiar ways) to construct new sets out of old sets. Only those things are sets.

A class is given by a decision procedure that, given a thing, determines whether that thing is a member of the class. All sets are classes because the procedure "verify that this is a member of this set" produces a class that is identical to the set. However, there are decision procedures that do not produce sets because the object they produce cannot be constructed by the axioms. Consequently, there are classes that are not sets. Then, the class is equivalent to its decision procedure -- i.e., each class is equivalent to the property expressed by its decision procedure. So a class, its decision procedure, and the property it encodes are all equivalent.

Eric Towers
  • 67,037