0
$\begingroup$

I have come across the prove that [Zorn's Lemma ==> AC] but am confused about the central statement, namely that we can take a set of all choice functions on subsets of X (lets just call it X, I mean the collection of sets we want to prove AC on).

If what we are trying to show is the existence of a choice function on our set X, how can we assume the existence of such a function for subsets of X? References would also be much appreciated

This is the Set of which I cannot prove the existence

The original proof is in German, but this is the link https://de.wikibooks.org/wiki/Beweisarchiv:_Mengenlehre:_Lemma_von_Zorn

$\endgroup$
7
  • $\begingroup$ At some point, in order to apply Zorn, you'll have to prove that the set of choice functions on subsets of $X$ is non-empty, i.e. that there is a subset of $X$ on which a choice function is defined. For instance, a singleton. $\endgroup$ Commented Jun 11 at 20:24
  • $\begingroup$ You can always define the set of choice functions on $X$. The axiom of choice says that this set is always inhabited (for nonempty $X$). $\endgroup$ Commented Jun 11 at 20:24
  • $\begingroup$ This proof is never given in any of the books I have read + I cant see how to do that without the axiom of choice. A subset of X is just another collection, and then the statement is equivalent to AC $\endgroup$ Commented Jun 11 at 20:25
  • $\begingroup$ You should edit your question with an actual proof and the precise step you're having trouble understanding. $\endgroup$ Commented Jun 11 at 20:26
  • $\begingroup$ That set $F$ exists by the axiom of separation. $\endgroup$ Commented Jun 11 at 20:34

3 Answers 3

2
$\begingroup$

You can always choose from a single non-empty set. This requires only Existential Instantiation, which is a rule of first-order logic. So, if $A$ is a non-empty set, that means that $\exists x(x\in A)$ is a true statement, and therefore we can add a new constant symbol $a$ and the axiom $a\in A$. Therefore, $\{\langle A,a\rangle\}$ is a choice function from the family $\{A\}$.

So, certainly, every singleton subset of $X$ admits a choice function. The same argument actually shows that given any $Y\subseteq X$ that admits a choice function and $A\in X$, $Y\cup\{A\}$ also admits a choice function: either $A\in Y$ in which case $Y=Y\cup\{A\}$, or else use Existential Instantiation to get a choice function on $Y$ and an element of $A$ and do as above.

So, if there was a maximal choice function, which is what Zorn's Lemma provides us with, it has to be defined on the entirety of $X$, otherwise, by the above argument, its domain can be extended, contradicting the maximality claim.

$\endgroup$
3
  • $\begingroup$ Okay, holy smokes that was quick and very insightful, but maybe for future readers let me see if I got your anwer right: The point is not to argue that this set of all choice functions already contains the choice function on X, but rather that: 1. it is not empty 2. all chains in it satisfy Zorns Lemma. $\endgroup$ Commented Jun 11 at 20:37
  • $\begingroup$ Yeah. And while technically you don't really need to argue that it's not empty, since the empty set itself is a choice function from itself, it is more illuminating to separate the non-empty case. $\endgroup$
    – Asaf Karagila
    Commented Jun 11 at 20:39
  • 1
    $\begingroup$ @CopperCableIsolator The nonemptiness is a consequence of "every chain has an upper bound" (by taking the empty chain). So in order to establish that the set satisfies Zorn's Lemma, one often needs to establish non-emptiness separately, as the general argument for upper bounds often does not work for the empty chain. (Here it does, since the upper bound for a chain of functions is the union of the functions with the union of the domains, which in this case gives the empty function on the empty set, which is a "partial choice function".) $\endgroup$ Commented Jun 11 at 20:55
1
$\begingroup$

It often happens that we can specify a certain property without being able to prove that anything has that property.

So if $U$ is a collection of subsets of a set $X$, we can say that a choice function $c$ for $U$ is a function $U \to X$ such that for all $A \in U$, $f(A) \in A$. Formally, the set $C$ of all choice functions for $U$ can be defined thus: $$ C = \{c: U \to X \mid \forall A \in U. c(A) \in A\} $$

My $C$ is the set $F$ in your linked image if you unwind the definition of $U \to X$ (the set of functions from $U$ to $X$). It is can be seen to be non-empty by the axiom of comprehension.

The set $C$ may or not be empty: it certainly is empty if $\emptyset \in U$. It can be proved without using $AC$ not to be empty if $\emptyset \not\in U$ and $U$ is finite. AC says that it is always non-empty if $\emptyset \not\in U$.

$\endgroup$
0
$\begingroup$

At the beginning of the proof you don't know that such a choice function exists for all subsets of X, but you know that it at least exists for some (for example, even without the axiom of choice you can prove easily that, there is a choice function for all finite subsets of X). Using the fact that choice functions exist on some subsets and that these choice functions satisfy Zorn's Lemma, you get a choice function for X (and thus, actually all subsets of X have choice functions, but you only know this after the proof).

$\endgroup$

You must log in to answer this question.

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