7
$\begingroup$

I am quite sure the following fact must have been known for set theorists, though I could not find it anywhere.

If $r$ is random over $L$ and $x\in L[r]\setminus L$, then there must be some real $r_0$ random over $L$ so that $r_0\in L[x]$ and $x\in L[r_0]$.

So where I could find it?

Addition

Monroe Eskew has given an answer to the question. I would like to explain the phenomenon from a recursion theory point of view.

As Jason pointed out, the result is corresponded to Demuth's theorem in recursion theory. Actually it can be proved by applying Demuth's argument.

Suppose that $r$ is random and $x\in L[r]\setminus L$. Then there is a condition $r\in p\Vdash x\mbox{ is a real}$. Since random forcing is c.c.c., we may in $L$ find a sequence $\{p^i_n\}_{i,n\in \omega}$ stronger than $p$ so that for every $n$, $\{p^i_n\}_{i\in \omega}$ is a maximal antichain below $p$ and each $p^i_n$ decides a value of $x$. Coding these condition and the relation $p^i_n\Vdash x(n)=j_i$ into a single real $z\in L$. Then $x\leq_T r_0\oplus z$ with a use function $f\in L[r]$. Then $f$ is dominated by a function $g\in L$. Now using these facts and applying Demuth's argument, we may obtain a real $r_0$ which is also $L$-random so that $x\oplus z\oplus g\equiv_T r_0\oplus z\oplus g$.

So $x\in L[r_0]$ and $r_0\in L[x]$.

$\endgroup$
8
  • 1
    $\begingroup$ What exactly do you mean by "random over $L$"? Is it "$(L,\mathbb P)$-generic" where $\mathbb P$ is Random (Solovay) forcing and $L$ is the constructible universe? $\endgroup$ May 26, 2014 at 11:58
  • 3
    $\begingroup$ Since you added the computability theory tag I can't help pointing out the connection (as I see it) to algorithmic randomness: Your question seems to be a set theoretic version of the following computability theoretic statement: "If $r$ is Martin-Löf random and $0<x\leq r$, then there exists a Martin-Löf random $r_0$ so that $r_0 \equiv x$." If $\leq$ is truth-table reduction and $\equiv$ is Turing equivalence, this is a theorem of Demuth. $\endgroup$
    – Jason Rute
    May 26, 2014 at 16:02
  • 1
    $\begingroup$ I changed the title to something more specific; I am not sure it is chosen well though and you could change it using "edit". In any case please try to avoid extremely general titles like "Reference request"; some places the only thing of a question that is shown is its title and this thus should convey something somewhat specific about the question. $\endgroup$
    – user9072
    May 26, 2014 at 16:08
  • 1
    $\begingroup$ I was just kidding. Thanks for pointing out this. $\endgroup$
    – 喻 良
    May 26, 2014 at 16:14
  • 2
    $\begingroup$ @quid I made the title even more descriptive (hopefully) $\endgroup$ May 26, 2014 at 16:19

1 Answer 1

9
$\begingroup$

I'm not sure if this specific claim is stated explicitly anywhere, but it follows from the more general discussion about intermediate extensions in Jech, page 247. If we take any $x \in L[r]$, then there is a complete subalgebra $A$ of the random algebra associated to $\dot{x}$, with the property that $L[x]=L[G_r \cap A]$. But any complete subalgebra of a measure algebra is also a measure algebra. So $A$ is isomorphic to random forcing.

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.