Let $X$ be an arbitrary infinite set, can we always find a bijective map $T: X\rightarrow X$ such that for any finite (nonempty) subset $F\subset X$, $T(F)\neq F$ ? This question is related to another post.
Asked
Active
Viewed 137 times
2
-
Are there any assumptions of choice in the theory? – Asaf Karagila Oct 09 '11 at 21:23
-
If $F=\emptyset$, then $T(\emptyset)=\emptyset$ for any $T$. – SL2 Oct 09 '11 at 21:26
-
No assumption in my mind. You may assume that the cardinality of $X$ $\le$ $c$. – Syang Chen Oct 09 '11 at 21:26
-
If $X$ is finite and $F=X$ then $T(F)=F$. – George Lowther Oct 09 '11 at 21:31
-
Sorry, corrected. – Syang Chen Oct 09 '11 at 21:32
-
The statement that $T(F)\not=F$ for any finite $F$ is equivalent to saying that the orbits of $T$ are infinite. Or, equivalently, $T^m(x)\not=T^n(x)$ for $m\not=n$ and $x\in X$. – George Lowther Oct 09 '11 at 21:44
1 Answers
4
This is easy enough when $X$ is finite or countably infinite. In the general case I think it requires the Axiom of Choice. Given AC we know $X\simeq X\times \mathbb Z$ for any infinite $X$, and $X\times\mathbb Z$ can be made to satisfy your property by letting $T$ shift each copy of $\mathbb Z$ one position to the right (or left).
hmakholm left over Monica
- 291,611
-
I don't see right away why this should be true for every infinite set, in fact I'm not even sure I see it for finite sets (it may just be the long day I had in thinking without the axiom of choice that is speaking, however I still can't see why all finite subsets must be moved.) – Asaf Karagila Oct 09 '11 at 21:27
-
@Asaf, it's a standard (but not trivial) consequence of AC that $\aleph_\alpha\times\aleph_\beta \simeq \aleph_{\max(\alpha,\beta)}$. – hmakholm left over Monica Oct 09 '11 at 21:32
-
Come on, I'd guess that you figured out that I'd know that by know. I don't see how this implies that there are no finite subsets which are "globally" fixed (which implies that no subset is being fixed) – Asaf Karagila Oct 09 '11 at 21:34
-
Great! Thank you! @Asfa, he translates every point in the interger index by $1$. – Syang Chen Oct 09 '11 at 21:35
-
@Asaf, if you have a finite nonempty $F\subset X\times\mathbb Z$, then there must be a least $n$ such that $(x,n)\in F$ for some $x$. But this $(x,n)$ cannot be a member of $T(F)$ because $T$ adds one to all of the $n$'s. – hmakholm left over Monica Oct 09 '11 at 21:38
-
@Henning: Yes, I knew that it was this long day of working without the axiom of choice that got me confused here! :-) – Asaf Karagila Oct 09 '11 at 21:42
-
@Asaf, yeah, that's why I assumed that it must have been the cardinal arithmetic you'd blanked out on (since that was the only use of choice in my argument). – hmakholm left over Monica Oct 09 '11 at 21:45