-1
$\begingroup$

N is the set of natural numbers, an infinite set of finite elements. A digit string is an ordered list of digits from the set {0,...,9}, such that no element starts with 0 except zero (this last for convenience). D is the set of all digit strings; it is infinite and contains both finite and infinite elements. How do I prove that an enumeration of D cannot contain every element of D, i.e that it can only be a partial enumeration? Ideally without using Cantor's diagonal argument.

Obviously if I enumerate the elements of N using strings from D as {0, 1, 2, 3, ...}, the enumeration will contain no infinite strings. But I could interleave finite strings with infinite ones in order to get around this.

  • 0
    Well, the irrationals between $0$ and $1$ each give rise to such a string...2017-01-04
  • 0
    It seems like you've constructed the countable numbers.Skip the interleaving.2018-06-21

1 Answers 1

-1

Why do you want to avoid Cantor's diagonal argument? This is a case where it would be the most natural way to do it. I assume you understand how to use it, just assume an enumeration of the set and construct a string by for each string in the enumeration append $1$ if the n:th digit of the n:th string is not $1$ (including the case where the n:th string is to short to have an n:th element). This string is not present in the enumeration.

Alternatively you can use the fact that the power set is always a larger set, but to prove that you'd use Cantor's diagonal argument.

A third option is to realize that your set has the cardinality of $\mathbb R$ which is known to be non-countable, but again you'd probably rely on Cantor's argument to prove that $\mathbb R$ is not countable.

A more elaborate method would be to for each sequence $d\in D$ map (by $f$) a real number $x$, by writing $0.$ followed by the digits of the sequence. Note that this mapping is not injective which is not a problem. This maps $D$ to a set $E$ (namely ${0}\cup(0.1,1]$). Now consider the set $X$ of single-element subsets of $E$. We have $|D|\ge |E| = |X|$. Now we assume that $X$ is countable, but then we have since $m(\xi) = 0$ for all $\xi\in X$ (where $m$ is the Lebesgue measure). If $X$ were countable we would have $m(E) = m(\bigcup_X \xi) = 0$, but we know that $m(E)=0.9$ which contradicts that $X$ is countable and since $D$ is no smaller it can't be countable. (I'm not sure if this relies on Cantor's argument, but I can't see any obvious such dependencies right now)

  • 0
    Thanks. It follows from this (last) argument that if you try to enumerate D, there must be elements in the set that will not be part of the enumeration. It therefore follows that you don't need Cantor's diagonal argument to show that R can't be enumerated using decimal representations. Which is what I wanted.2017-01-04
  • 1
    "but we know that $m(E)=0.9$" *How* do we know that? That is, how do we know that there isn't some magical way to cover $E$ with open intervals with total size $<0.9$? This requires work to prove, and will essentially subsume (something like) Cantor's argument (probably more like Cantor's original argument than his diagonal argument, actually).2017-01-12
  • 0
    @NoahSchweber Because $\int_{0.1}^1dx=0.9$2017-01-12