1
$\begingroup$

Let the set $A$ be finite and $\emptyset \notin A$. How can I, without using the axiom of choice, prove by mathematical induction that there exists a function $f : A \rightarrow \bigcup A$ satisfying $f(a) \in a$ for all $a \in A$?

$\endgroup$
1
  • $\begingroup$ @AlessandroCodenotti really, thank you, I fixed it) $\endgroup$
    – Andrew S.
    Commented May 7 at 9:33

1 Answer 1

4
$\begingroup$

It holds vacuously for $A = \emptyset$: then $\bigcup A = \emptyset$ and the empty function $\emptyset: \emptyset \to \emptyset$ trivially fulfills $f(a) \in A$ for all $a\in A$ -- because it is impossible to find a counterexample in an empty set.

Let $n_0 \in\mathbb{N}$ be the smallest number such that there is $A$ of cardinality $n_0$ with $\emptyset \notin A$, but no choice function $f:A \to \bigcup A$ with $f(a)\in a$ for all $a\in A$. We just saw that $n_0$ cannot be $0$.

So we have $n_0 > 0$. Let $a_0\in A$. Because $n_0$ is the smallest integer such that there is no choice function, there is a choice function $f:\big(A\setminus \{a_0\}\big)\to \bigcup\big(A\setminus\{a_0\}\big)$. Pick $x_0\in a_0$ (which is possible since $a_0\neq \emptyset$). Now set $$f' = f \cup \big\{(a_0, x_0)\big\}.$$ It is easy to verify that this is a choice function for $A$, contradicting the assumption that $A$ does not have a choice function.

$\endgroup$
9
  • 1
    $\begingroup$ Great answer, thank you so much!) $\endgroup$
    – Andrew S.
    Commented May 7 at 9:37
  • 2
    $\begingroup$ I think it doesn't matter for your last paragraph that $n_0 > 1$, as opposed to $n_0 > 0$. That is, you don't need to handle the $n_0 = 1$ case separately. $\endgroup$
    – LSpice
    Commented May 7 at 12:12
  • 1
    $\begingroup$ Thanks @LSpice - will use the shortcut you mention $\endgroup$ Commented May 7 at 12:45
  • 2
    $\begingroup$ You also really don’t need to phrase the proof as by contradiction. Anyways, the question itself is off topic here so it doesn’t need a perfect answer. $\endgroup$ Commented May 7 at 14:10
  • 1
    $\begingroup$ Horrible, horrible unecessary use of proof by contradition. Straight induction would have worked. $\endgroup$ Commented May 7 at 20:55

Not the answer you're looking for? Browse other questions tagged .