When I was first introduced to sets the common "description" in place of a formal definition used to be something like a set is a collection of objects that do not have to be related in anyway prior to the formation of the set though the often are in mathematics. And a rule of sets was that an element of a set can be represented once and only once e.g. the following is not a set due to the repeat representation of $2$ $$\{ 1, 2, 3, (4-2) \} $$. Suppose I wanted to create a set which contained the representations (denotation) of the number 2: $$\{2, (\frac{2}{1}), (\frac{4}{2}), 5-3, 2k -2(k-1),...\}$$

In the sense(i.e. not necessarily intended as the philosophical Fregean term sense) of the first example the collection above is not a set but in the sense that it is a collection of representations of $2$ it is.

So it would seem that whether something is a set depends on another property, in addition to being a collection and having each element represented once : that is, context.

What would be a better description(not necessarily a definition) of a set if we want to describe its context? We could use the idea of a universe but that itself is conceived of as a set so the universe cannot provide the context exactly can it? For example, I can see how the universe being $\mathbb N$ or $\mathbb R$ can be of help but the universe of $\mathbb N \cup \ $ (Set of Representations of $\mathbb N$) does not seem like it provides much clearity.

  • $\begingroup$ Please do not add the set-theory tag, since it does not apply. You are not discussing a specific formalism, but rather a vague notion of context which isn't really mathematical. $\endgroup$ – Andrés E. Caicedo Jun 20 '17 at 3:15

If you want to distinguish between a set of numbers and a set of representations of numbers, the most obvious thing would be to describe the set as a subset of some well-understood (or at least better-understood) set. For example,

The set $\{1,1,2, \frac{2}{1},3\}\subset\mathbb N$ is a set of natural numbers with three distinct elements. Let $A$ be the set of strings of ASCII characters. The set $\{1,\text{I},\text{one},\text{uno}\}\subset A$ is a set of representations of the number $1$, with four distinct elements.

This ties closely to the use-mention distinction, between using the number $1$ and mentioning the number $1$. We often use quotation marks in English to mention words, and this could also be applied to distinguish a number from its representation providing we were clear about it:

The set $\{1, \text{`$1$'}\}$ contains the number $1\in\mathbb N$ and the representation $\text{`$1$'}\in A$.

Of course, if you want to define the set $\mathbb N$ from scratch this may not be very useful. But if all you are concerned about is specifying how the listed elements of a set should be regarded, this ought to be sufficient.

  • $\begingroup$ Perhaps this is getting too philosophical but isn't the quotational '1' another representation ultimately? It is still a syntactical structure although I completely agree with you that intuitively it's enough of a distinction to work with. $\endgroup$ – Red Jun 18 '17 at 18:31
  • $\begingroup$ And I guess you're comment about defining $\mathbb N$ is about the fact of how we assign symbols to the sucessor function, that is that picking 1 to represent the use of the number and '1' to represent its representation is arbitrary? Because I don't see that as much of a problem. $\endgroup$ – Red Jun 18 '17 at 18:39

In the usual approach to numbers from (an empty) scratch, the von Neumann rank of a natural number $n$ is itself, that of a rational number is finite, while each real number defined as a Dedekind cut $(Q,Q')$ have von Neumann rank $\omega+1$. This means however that an identity like $\sqrt{2} \times \sqrt{2}=2$ has to be properly interpreted if we wish to view $2$ as a rational number. Thus context needs to be supplied if one is to interpret that identity correctly.

On the other hand, it seems curious that one can't make a distinction between $\mathbb R$ as a set of real numbers and $\mathbb R$ as an element of $\mathcal{P}(\mathbb R)$ as they are exactly identical in the von Neumann hierarchy. Here also context needs to be supplied to explain what you are doing.

All this leads me to a tentative conclusion that the concept of equality is in fact a vague one in modern mathematics. Sometimes this is handled by making a distinction between equality and equivalence, and further between natural and "un"natural isomorphisms, but these seem somewhat ad hoc.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.