3

Let $X$ be an infinite set. Then there exists a partition $\lbrace A, B \rbrace$ of $X$ such that $A, B, X$ are equinumerous.

Can you prove it in $\mathrm{ZFC}$?
Can you prove it in $\mathrm{ZF}$?

Stefan Mesken
  • 16,852
  • 3
  • 28
  • 48
Τίμων
  • 1,699

1 Answers1

4

We can prove it in $\operatorname{ZFC}$, but not in $\operatorname{ZF}$:

In $\operatorname{ZFC}$ there is some initial ordinal $\kappa$ and a bijecion $f \colon \kappa \to X$ and also some bijection $g \colon \kappa \times 2 \to \kappa$ (the function $g$ exists also if we drop choice, but $f$ may not exist). Let $h = f \circ g$, $A = \{ h(\xi, 0) \mid \xi < \kappa \}$ and $B = \{ h(\xi, 1) \mid \xi < \kappa \}$. Clearly $A \cong B \cong X \cong \kappa$.

On the other hand, if we drop choice, there may be some infinite, but Dedekind finite set $X$ and those sets don't even allow for a proper subset $A$ that is in bijection with $X$.

Stefan Mesken
  • 16,852
  • 3
  • 28
  • 48
  • 1
    It maybe of interest that the assumption that every infinite set $X$ can be partitioned into "$X$-many" sets of size $X$ already implies the axiom of choice - and there is an elementary proof for this. See Tarski's Theorem. – Stefan Mesken Mar 03 '16 at 14:45
  • 2
    It can get even worse - it is consistent with ZF that there is an infinite set which can't even be partitioned into two infinite pieces! (Such sets are called amorphous.) – Noah Schweber Mar 06 '16 at 01:04