4

I have seen these three symbols, $\aleph_0$, $\omega$ and $\mathbb{N}$, a lot in my reading (mostly in analysis, I have very limited experience in set theory). I have seen in various places they are used interchangeably, which is confusing for me.

There is no problem that the symbol $\mathbb{N}$ denotes the set of natural numbers. (By convention, the number $0$ may or may not be in the set.) The aleph null $\aleph_0$ is defined as the "cardinality" of the set $\mathbb{N}$. This Wikipedia article says that $\omega$ is the first infinite ordinal. I have seen people use $\mathbb{R}^\omega$ for the set of all real sequences (see, for instance, Munkres's Topology); some people use $\mathbb{R}^{\mathbb{N}}$ instead, which suggests that $\omega$ and $\mathbb{N}$ may be the "same" in some sense. On the other hand, I have never seen $\mathbb{R}^{\aleph_0}$.

The definitions of these three concepts are quite different, yet they seem to be closely related.

So my question is: how exactly are they related to each other and in what sense they are (possibly) the same?

Gerry Myerson
  • 179,216
  • 1
  • 6
    For me $\Bbb N=\omega=\omega_0=\aleph_0$, since I take cardinals to be initial ordinals. Using $\aleph_0$ emphasizes that one is talking about it as a cardinal, though in general I prefer to use $\omega$ whether I’m using it as an ordinal or as a cardinal, just as I generally prefer to use $\omega_1$ (rather than $\aleph_1$) for the first uncountable cardinal. Context generally suffices to make it clear which sense is intended. – Brian M. Scott Apr 18 '20 at 17:50
  • 1
    @Brian: Working in set theory, I can tell you that often context is not 100% clear. Even more so, you don't have "fifth" fingers on your hand, you have "five" fingers on your hand (assuming a healthy hand, anyway). Linguistically it makes sense to make the distinction between ordinals and cardinals. – Asaf Karagila Apr 18 '20 at 20:53
  • 2
    To the OP: Asaf's comment notwithstanding, ignoring pedagogy $\mathbb{N}=\omega=\omega_0=\aleph_0$ is literally true. The real ambiguity comes in when we use the same symbols for ordinal and cardinal operations: e.g. "$\omega+\omega\not=\omega$" is true but "$\aleph_0+\aleph_0\not=\aleph_0$" is false. – Noah Schweber Apr 18 '20 at 22:18
  • @Noah: I did not dispute that fact. – Asaf Karagila Apr 18 '20 at 22:19
  • @AsafKaragila Sorry, I didn't mean to imply that - I just wanted to clarify for the OP. (I've edited to make that clear.) Another comment to the OP: if you're familiar with computer programming, it may help to think of the difference between the floating-point number $1.0$ and the integer $1$. – Noah Schweber Apr 18 '20 at 22:19
  • @BrianM.Scott: Thanks for the comment. What is $\omega_0$? Presumably, is it completely the same as $\omega$ with respect to definitions? –  Apr 18 '20 at 22:30
  • @NoahSchweber: thanks a lot. So your previous comment means that $\mathbb{N}=\omega=\omega_0=\aleph_0$ as sets while the difference lies in the way one does "arithmetic" (whenever defined), correct? –  Apr 18 '20 at 22:35
  • @Mars Basically, yes. There's no difference between the objects, but we've overloaded some related notation (the opposite situation: the same symbol standing for differnet objects!) in such a way that the symbols here give important context clues. E.g. "$\omega$" tells us that any nearby addition symbols probably mean ordinal addition, and so on. – Noah Schweber Apr 18 '20 at 22:44
  • 3
    @AsafKaragila: When I was actively doing mathematics, I was mostly working in set-theoretic topology, and I generally did find it clear. And I continue to prefer, e.g., $\omega_1$ for both the ordinal and the cardinal. For what it’s worth, I’m fairly sure that I picked up most of my preferences from Ken Kunen. – Brian M. Scott Apr 18 '20 at 23:17
  • "Taking cardinals to be initial ordinals" is assuming the Axiom of Choice (that all sets can be well ordered). – GEdgar Apr 19 '20 at 00:04
  • I read both the tag information for elementary-set-theory and set-theory before asking the question. Also, I have read this question regarding the difference between these two tags. While I do consider my question should be "elementary" to set theorists, –  Apr 19 '20 at 01:33
  • I'm not sure if the tag elementary-set-theory is appropriate here: Topics include intersections and unions, differences and complements, De Morgan's laws, Venn diagrams, relations and functions, countability and uncountability, etc. More advanced topics should use (set-theory) instead. But I will not argue with the editor for the moment. –  Apr 19 '20 at 01:33
  • @NoahSchweber: Thanks for that. I'm a bit confused by your comment with floating-point numbers and integers. How are they related to the relation among $\mathbb{N}$, $\omega$ and $\aleph_0$? –  Apr 19 '20 at 01:36
  • @GEdgar: where is that quote from? –  Apr 19 '20 at 01:38
  • @Mars (In case you don't get a reply from GEdgar) I think they are referring to Brian Scott's first comment where he says he takes cardinals to be initial ordinals. – spaceisdarkgreen Apr 19 '20 at 02:58
  • I don't generally like illustrating mathematics using computing examples but this is a good one. The point here is that many computing languages overload symbols and their exact meaning depends on the types of the operands. E.g. In some languages $1 / 3 = 0$ but $1.0 / 3.0 = 0.333...$. $+$ is even commonly for the very different string concatenation: "bad" + "john" = "badjohn". – badjohn Apr 19 '20 at 06:51
  • @Mars Yes, the quote is from the comment by Brian M Scott – GEdgar Apr 19 '20 at 13:12

3 Answers3

4

It may be useful to separate the facts that should be true in any reasonable set-theoretic foundation from the facts that are true by convention in the usual foundation.

Generally true: $\aleph_0$ is the cardinal number of a countably infinite set. $\omega=\omega_0$ is the order-type of a simple infinite sequence (an infinite sequence in which each element has only finitely many predecessors). $\mathbb N$ is the set of natural numbers.

Convention 1 (von Neumann): Any ordinal (= order-type of a well-ordered set) is identified with the set of strictly smaller ordinals. Thus, $0$ is the empty set, $1=\{0\}$, $2=\{0,1\}$, etc., and $\omega=\{0,1,2,3,\dots\}$.

Convention 2: A cardinal number is identified with the smallest ordinal of that cardinality. Thus, $\aleph_0=\omega$. (This convention depends on the axiom of choice in general, to ensure that every cardinality is the cardinality of some ordinal. But this is not an issue for $\aleph_0$, which is the cardinality of $\omega$.)

Convention 3: $0$ is a natural number. (A nontrivial number of respectable mathematicians disagree with this and start the natural numbers with $1$.) So $\mathbb N=\omega$.

In the end, if you adopt all these conventions, you have $\aleph_0=\omega_0=\omega=\mathbb N$. If you adopt other conventions (or no conventions), you need to check what they say about these things, but the general facts that I listed first should still be true.

Andreas Blass
  • 71,833
2

What you are actually encountering here is implicit typing in mathematics, which is prevalent in almost every area of mathematics but is rarely taught in mathematical pedagogy. $\mathbb{N}$ is the type of natural numbers, and if you have members of $\mathbb{N}$ then all you can do on them are operations that require natural numbers as inputs. For example we define various exponentiation operations:

$S^n$ is the set of $n$-tuples from $S$, for any set $S$ and $n∈\mathbb{N}$.

$x^0 = 1_M$ and $x^{n+1} = x^n·x$ for any monoid $(M,·,1_M)$ and $x∈M$ and $n∈\mathbb{N}$.

$x^y = \exp(y·\ln(x))$ for any $x,y∈\mathbb{R}$ such that $x>0$.

$S^T$ is the cardinality of the set of functions from $T$ to $S$, for any sets $S,T$.

$k^m$ is the cardinality of the set of functions from $m$ to $k$, for any cardinals $k,m$.

$k^\varnothing=\{\varnothing\}$ and $k^m = \sup\{ k^p·q : p∈m ∧ q∈k \} $, for any ordinals $k,m$.

Implementation in set theory is very much irrelevant to the intrinsic mathematics. For example, $\mathbb{N}$ is implemented by $ω$ in modern set theory based on ZFC, and hence $0$ is implemented by $\varnothing$, but we never think of $0^2$ as $\varnothing^2$. Why? Because these operations are overloaded but disambiguated by the implicit input types. $0$ has implicit type $\mathbb{N}$, while $\varnothing$ has implicit type "set".

Types are implicitly introduced by mathematical definitions. For example, $\aleph_k$ is defined as the $k$-th cardinal, and this implicitly defines $\aleph_k$ to have implicit type "cardinal". In contrast, $ω$ is defined as the first infinite ordinal, so its implicit type is "ordinal". That is why $ω^ω$ is a countable ordinal while ${\aleph_0}^{\aleph_0}$ is an uncountable cardinal, despite $\aleph_0$ often being implemented as $ω$ in modern set theory.

Your three examples would all be understood by mathematicians who know the individual terms, due to implicit type coercion. That is, when none of the (overloaded) defined operations have matching type signature, we would pick the closest one that is compatible with the inputs according to their actual implementations. Since $\aleph_0$ is a cardinal and cardinals are usually implemented as ordinals, and ordinals are sets at the bottom of it all, we have an available implicit type coercion of $\aleph_0$ from "cardinal" to "ordinal" to "set".

Specifically, $\mathbb{R}^{\aleph_0}$ would be automatically interpreted as exponentiation of sets, which yields the set of functions from $\aleph_0$ to $\mathbb{R}$. $\mathbb{R}^\mathbb{N}$ needs no type coercion. $\mathbb{R}^ω$ yields the set of functions from $ω$ to $\mathbb{R}$, which is of course no different from $\mathbb{R}^{\aleph_0}$ at the implementation level, but at the communication level conveys that the input is an ordinal index, which corresponds to the notion that an ordinary infinite sequence of reals is an $ω$-sequence. Recursive definition of sequences only works for indices from a well-ordering and ordinals are canonical well-orderings (see this post), so it should not be surprising to see ordinals used as exponents to indicate the length of sequences from a set. Another example along this line is $S^{<ω}$, which denotes the set of all finite sequences from $S$, where "$<ω$" is suggestive of the meaning since ordinals less than $ω$ are finite.

user21820
  • 57,693
  • 9
  • 98
  • 256
1

Two sets $A$ and $B$ have the same cardinality iff there exists a bijection $f : A \rightarrow B$.

Two sets $A$ and $B$ have the same order type (under orders $\leq_A$ and $\leq_B$, respectively) iff there exists a bijection $f : A \rightarrow B$ such that $x \leq_A y \longleftrightarrow f(x) \leq_B f(y)$ for all $x, y \in A$.

An ordinal is a hereditarily well-founded and hereditarily transitive set. In ZF, we define the order type of a set as the (unique) ordinal that has the same order type under $\in$.

In ZFC, the von Neumann cardinal assignment defines the cardinality of a set as the smallest ordinal that has the same cardinality. This is called the initial ordinal of that cardinality, or cardinal. Thus $\omega_0$ and $\omega_0+1$, which are different ordinals, have the same cardinality.

$\omega_0$ is the smallest infinite ordinal, i.e. the order type of $\mathbb{N}$.

$\aleph_0$ is the smallest infinite cardinal, i.e. the cardinality of $\mathbb{N}$.

Under the von Neumann cardinal assignment, $\aleph_0$ is defined as $\omega_0$. However, using the symbol $\omega_0$ or $\aleph_0$ indicates whether we are treating it as an ordinal or as a cardinal, respectively. For example, $\omega_\alpha + 1$ has a different order type than $\omega_\alpha$, so we say that $$\omega_\alpha + 1 \neq \omega_\alpha$$ but they have the same cardinality, so we say that $$|\omega_\alpha| + 1 = \aleph_\alpha + 1 = \aleph_\alpha = |\omega_\alpha|$$

$\omega_0$ is the set of all finite ordinals. We notice that these finite ordinals under ordinal arithmetic behave like the natural numbers $\mathbb{N}$ under natural-number arithmetic. That is, we have an isomorphism $\mathbb{N} \cong \omega_0$.

$A^B$ is the set of functions $B \rightarrow A$. Thus $\mathbb{R}^{\omega_0}$ denotes the set of functions $\omega_0 \rightarrow \mathbb{R}$. Because $\mathbb{N} \cong \omega_0$, $\mathbb{R}^\mathbb{N} \cong \mathbb{R}^{\omega_0}$. That is, they are isomorphic.

See also this question.

user76284
  • 5,967