6

I just need help trying to create a proof that shows that an infinite set has a countable subset. Is it as simple as taking arbitrary values of the finite set and listing them in their own subset?

CopyPasteIt
  • 11,366
a.nas
  • 117
  • 1
    you mean an infinite countable subset? – TheEsnSiavashi Sep 29 '17 at 18:19
  • 1
    If you select an element $x_1$ from the infinite set $S$ does the set $S-{x_1}$ contain an element? If so can you pick an element $x_2$ from that set? Does the set $S-{x_1,x_2}$ contain an element? Etc. – John Wayland Bales Sep 29 '17 at 18:20
  • 5
    Note that any such proof will explicitly need the axiom of choice: https://en.wikipedia.org/wiki/Amorphous_set – Steven Stadnicki Sep 29 '17 at 18:22
  • @StevenStadnicki Indeed! Or does it? Is the Axiom of Choice needed to prove that an infinite set minus a finite set is infinite? – John Wayland Bales Sep 29 '17 at 18:22
  • Set $S_0=S$, and given an infinite set $S_n$ pick an element $x_n$ out of $S_n$ and set $S_{n+1}=S_n\setminus{x_n}$. You need to resolve a couple of things: (1) What allows to pick an element from a non-empty set? (2) Does this algorithm run without halting? (E.g. Does it guarantee that each $S_n$ is infinite?) – Sangchul Lee Sep 29 '17 at 18:28
  • Countable means not uncountable. That is, countable means "finite or countably infinite". Anyone who objects to this usage will have to either say that finite sets are uncountable or that "uncountable" and "not countable" are different things. – DanielWainfleet Aug 11 '18 at 08:09
  • The empty set is countable. It is also subset of every set, including the uncountable sets. $\square$ – celtschk Sep 12 '18 at 10:25
  • "Countable" is often used as a shorthand for "countably infinite". (And if I define an uncountable set as a set with strictly greater cardinality than natural numbers, then in absence of axiom of choice this may not be the same as "not countable" - neither finite nor countably infinite. If disambiguation is needed, I can call sets that satisfy my definition as "strictly uncountable".) – Mike Rosoft Mar 13 '20 at 06:37

4 Answers4

7

Definition: The statement that a set $S$ is infinite means that if $N$ is a natural number then $S$ contains $N$ distinct elements.

[Note: If an infinite set is defined in this way, then it automatically follows that an infinite set minus a finite set is infinite.]

Suppose $S$ is infinite.

Since $1$ is a natural number, $S$ contains an element $x_1$.

Since $2$ is a natural number, $S$ contains an element $x_2$ distinct from $x_1$.

So there is a two-element subset $U_2=\{x_1,x_2\}$ of distinct elements of $S$.

For each $N\in\mathbb{N}$, $S$ contains an element distinct from each element in $U_N=\{x_1,x_2,\cdots,x_n\}$, so define $U_{N+1}=U_N\cup\{x_{N+1}\}$ where $x_{N+1}$ is an element of $S$ distinct from each element of $U_N$.

Let $$U=\bigcup_{N\in\mathbb{N}}U_N$$

Then $U$ is a countable subset of $S$.

ADDENDUM There is an issue which I glossed over when making this argument.

For each $N$ we know that there is a subset $V$ of $S$ containing $N+1$ elements of $S$. So $V$ contains an element, call it $x_{N+1}$, which is distinct from each element of $U_N=\left\{x_n,x_2,\cdots,x_N\right\}$. Let $U_{N+1}=\left\{x_n,x_2,\cdots,x_N,x_{N+1}\right\}$.

  • 2
    "Definition: The statement that a set S is infinite means that if N is a natural number then S contains N distinct elements." Um... what? So $\mathbb R$ is infinite. 27 is a natural number. So there are 27 distinct real numbers. Or did you mean that it will have 27 among others elements? If so, I've never seen this definition and it incorporates the result to be proven within it. – fleablood Sep 30 '17 at 01:34
  • @fleablood This is the definition of infinite set which learned in undergraduate school in the early 1960s. Keep in mind that, for example, a set that contains five elements contains four elements. – John Wayland Bales Sep 30 '17 at 01:36
  • @fleablood The statement that the set $S$ is finite means that there is a natural number $N$ such that $S$ contains $N$ elements and such that $S$ does not contain $N+1$ elements. – John Wayland Bales Sep 30 '17 at 01:40
  • 6
    Wouldn't most texts define finite as there existing a bijection between {1.....n} and the set, (or have n elements) and define infinite as not finite. From which there existing finite subsets of an size would have to be proven. Which is easy but not a given. – fleablood Sep 30 '17 at 05:20
  • @fleablood You may be right, but does it matter? – John Wayland Bales Sep 30 '17 at 06:05
  • Welllll, I just think the idea of fundamental proofs at this level (Of course, infinite sets have countable subsets--- just pick out arbitrary elements, one at a time) is to hone our definitions and ability to reason via bijections and to set up models for questions properly. I worry these casual reasonings may be circular. They may not be. – fleablood Sep 30 '17 at 06:42
  • 1
    @fleablood You may be right. I learned these definitions from R. L. Moore at the University of Texas who was somewhat of an iconoclast. Then I spent most of my career teaching undergraduate engineering majors, so I never had to revisit the issue. – John Wayland Bales Sep 30 '17 at 06:52
  • @fleablood See my attempt to make sense of this... – CopyPasteIt Sep 12 '18 at 09:22
  • @fleablood: I prefer the definition as given in the answer, as it doesn't presuppose that all finite sets have an integer number of elements (you can, of course, prove it from that definition). And it is quite intuitive that if you can find any integer number of elements in it, the set must be infinite. – celtschk Sep 12 '18 at 10:32
  • 2
    Correct me if I am wrong, but I believe that the above proof implicitly makes use of axiom of dependent choice, which is an overkill. (The theorem is a consequence of axiom of countable choice.) – Mike Rosoft Mar 12 '20 at 23:31
  • @MikeRosoft You are right. This proof uses dependent choice, which is stronger than countable choice. – J. De Ro May 11 '20 at 21:15
2

Looking over the comments between John Wayland Bales and fleablood, I thought I would take a stab at this 'iconoclast approach', working in the realm of intuitive set theory.

Definition: A set is $S$ finite if there exist a natural number $n \gt 0$ such that every mapping $f: \{1, 2, \dots, n\} \to S$, the mapping $f$ is not an injection.

For a finite set $S$ we can consider the smallest $n_0$ where there are no injections. We then can say that $S$ has $n_0 - 1$ elements.

Proposition 1: If a finite set $S$ has $m$ elements then $S$ is equinumerous with $\{1,2,\dots,m\}$.
Proof
Let $g$ be an injection of $\{1,2,\dots,m\}$ into $S$. This mapping is necessarily a surjection. Suppose, to get a contradiction, that an element $s_0 \in S$ is not in the range of $g$. Then we can extend $g$ to another injective function by defining $g(m+1) = s_0$, but this is absurd. $\quad \blacksquare$

A set $S$ is infinite if it is not finite. This means that for every $n \gt 0$ there exist an injective mapping from $\{1,2,\dots,n\}$ into $S$.

Considering Dan Christensen's answer to the OP's question, we introduce the following axiom:

A set $S$ is infinite if and only there exist a function $f: S \to S$ that is an injection but is not a surjection.

CopyPasteIt
  • 11,366
  • In terms of epistemology (theory of knowledge), not of mathematics, per se, it seems wrong to define less abstract concepts in terms of more abstract concepts. Counting and natural numbers are less abstract than injections, surjections, bijections, etc. So to define a concept such as finite in terms of such higher abstractions rather than directly in terms of natural numbers seems somehow inappropriate. – John Wayland Bales Sep 12 '18 at 19:16
  • The statement that a set is finite means that there is a natural number $N$ such that $S$ does not contain $N$ elements. The statement that a set $S$ is infinite means that $S$ is not finite. What could be simpler? – John Wayland Bales Sep 12 '18 at 19:20
  • I would describe a set equinumerous with ${1,2,3,\cdots,m}$ as containing exactly $m$ elements or only $m$ elements. The statement that $S$ contains only $m$ elements means that $S$ contains $m$ elements but $S$ does not contain $m+1$ elements. – John Wayland Bales Sep 12 '18 at 19:58
  • @JohnWaylandBales I was negating your definition of infinite to define finite, and then proposition 1 was supposed to give us all a 'warm and fuzzy' feeling. Apparently I 'missed the boat'! – CopyPasteIt Sep 12 '18 at 23:15
  • The negation of "set $S$ is infinite" does not require any discussion of functions. A set $S$ fails to be infinite if there exists a natural number $N$ such that $S$ does not contain $N$ elements. It doesn't have to be complicated. – John Wayland Bales Sep 12 '18 at 23:23
  • @JohnWaylandBales Well, what does it mean to say that a set $S$ contains at least $N$ elements if not that there is an injection $F:{1,\dots, N}\to S$? – Gleison Stanlley Aug 22 '22 at 15:32
  • @JohnWaylandBales Also, why would "counting and natural numbers are less abstract than injections, surjections, bijections, etc." be true? How do you compare levels of abstractions? – Gleison Stanlley Aug 22 '22 at 15:34
  • @GleisonStanlley Which concepts does a child learn first? Numbers and counting? Or injections? Is there an epistemological reason for that order? – John Wayland Bales Aug 22 '22 at 15:40
  • @JohnWaylandBales When children count, what are they doing if not making bijections? The fact that they don't know how to formally define bijections doesn't mean they are not using it. Natural numbers are identified with the finite von-Neumann ordinals in ZFC by the way. Though for a non set-theorist Peano axioms are enough, children don't know them either. In order to prove things some rigor is needed. To prove that $2+2=4$ to a child you show your fingers, but in order to do Mathematics you need a stronger (and more rigorous) framework like PA. – Gleison Stanlley Aug 22 '22 at 15:51
  • @GleisonStanlley The fact that we can, once we develop the higher level abstraction 'bijection', identify the fact that that is what we are doing when we are counting does not mean that we did not know what counting was until we developed the concept of 'bijection.' In mathematics, we build systems of knowledge. The 'gold standard' of such systems are axiomatic systems. In an axiomatic framework we may find it useful to redefine concepts which, outside that framework, were defined ostensively. For example, redefining counting as a special case of a bijection. – John Wayland Bales Aug 22 '22 at 16:07
  • @GleisonStanlley But such specialized definitions do not deny or disparage definitions formed at lower levels of abstraction. – John Wayland Bales Aug 22 '22 at 16:07
  • @JohnWaylandBales You're putting "bijection" as a more abstract concept than that of counting. I don't think that the definition of counting as some sort of bijection is a "redefinition" according to some axiomatic framework: I think counting is some sort of bijection. You don't need to be working in an axiomatic set theory in order to understand the concept of function: this concept seems to rely on the same level of abstraction as that of the natural numbers. – Gleison Stanlley Aug 22 '22 at 17:47
  • @JohnWaylandBales Anyway, if someone asked me what "the set S contains $N$ elements" means, I don't know what else to answer if not to say that there is a bijection $F:{1,\dots, N}\to S$. – Gleison Stanlley Aug 22 '22 at 17:47
1

Definition: A set is infinite, if it can't be mapped one-to-one with an n-element set for any natural number n.

Lemma (can be proven using the principle of induction): If a set is infinite, then it has an n-element subset for every natural number n.

Proof:

  • Let $X$ be any infinite set.
  • By axiom of separation on $\omega \times P(P(X))$, there exists a function $f$ from non-zero natural numbers, where $f(n)$ is the set of all n-element subsets of $X$. (By lemma, $f(n)$ is non-empty for every n.)
  • By axiom of countable choice, there exists a sequence of sets $A_1, A_2, \dots$, such that $A_n$ is an n-element subset of $X$.
  • Let's define the following sequence of sets: $B_0=A_1$, $B_1=A_2 \setminus A_1$, $B_2=A_4 \setminus A_2 \setminus A_1$, ...; $B_n$ is the set of all elements of $A_{2^n}$, which are not an element of any $A_{2^m}$ where $m<n$ (any of $A_1$, $A_2$, $A_4$, ..., $A_{2^{n-1}}$). Observe that all these sets are finite, disjoint, and non-empty (the set $B_n$ has excluded at most $2^{n}-1$ different elements from $A_{2^n}$).
  • By axiom of countable choice, there exists a set containing a single element from each $B_n$. This set is a countably infinite subset of $X$. QED.
  • I would have though there would be an easier way, but I can't find one. Thanks a lot for the explicit proof. – user12861 Mar 17 '22 at 17:37
0

Let A be finite and non-empty then it is countable, by definition. If A is infinite then pick $a_1 \in A$. Then $A \setminus \{a_1\}$ is infinite. Pick $a_2 \in A \setminus \{a_1\}$. Then $A \setminus \{ a_1,a_2\}$ is infinite. Then the for each positive integer i, choose $a_i \in A \setminus \{ a_1, \dots ,a_{i-1} \}$. Hence the set S, defined by $a_i \in S$ $\iff$ i is a positive integer, is a countably infinite set hence countable.

In summary, you choose distinct points in the infinite set and form a set that consists of these points. Note that any set that consists of distinct points only is countable as you can index the terms by the positive integers and hence define a trivial bijection from the set to the positive integers.

  • 2
    You can't do this, unless you are assuming some form of AC. You can't induct infinitely. – Rushabh Mehta Jan 25 '20 at 22:03
  • Yes, I'm assuming axiom of choice. Nowhere did I mention induction. –  Jan 26 '20 at 14:23
  • 1
    The above is the same proof as from John Wayland Bales's answer, but not as well-written. (So - I believe - it implicitly uses axiom of dependent choice, but the theorem is a consequence of the weaker axiom of countable choice.) – Mike Rosoft Mar 13 '20 at 06:10