MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I don't known the correct credit for the following: every non-empty $\Pi^1_1$ set of reals contains some $X \in L_{\alpha}$ for some $X$-recursive $\alpha$. (Addison-Kondo?)

So, my question is: what is the least $\beta$, s.t. $\mathcal{P}(\omega) \cap L_\beta$ is a basis for all non-empty $\Pi^1_1$ sets?

share|cite|improve this question
up vote 5 down vote accepted

I can't access the full paper at the moment, but I'm pretty sure this is exactly the question addressed in the paper "A Note on the Kondo-Addison Theorem" by D. Guaspari

share|cite|improve this answer
Available online at – Trevor Wilson Jul 21 '12 at 14:15
Thanks a lot! The last theorem on obtainable ordinals is quite interesting. – Wei Wang Jul 22 '12 at 11:06

The least such ordinal is the least ordinal which cannot be a $\Delta^1_2$-well-ordering over natural numbers.

Let $$\delta^1_2=\mbox{ supremum of the }\Delta^1_2 \mbox{ wellorderings of } \omega,$$ and $$\delta=\min\{\alpha\mid L\setminus L_{\alpha}\mbox{ contains no }\Pi^1_1 \mbox{ singleton}\}.$$

We claim that $\delta=\delta^1_2$.

$\mathbf{Proof}$: If $\alpha<\delta$, then there is a $\Pi^1_1$ singleton $x \in L_{\delta}\setminus L_{\alpha}$. Since $x\in L_{\omega_1^x}$ and $\omega_1^x$ is a $\Pi^1_1(x)$-wellordering, it must be that $\alpha<\omega_1^x<\delta^1_2$. So $\delta\leq \delta^1_2$.

If $\alpha<\delta^1_2$, there is a $\Delta^1_2$ wellordering relation $R\subseteq \omega\times \omega$ of order type $\alpha$. So there are two arithmetical relations $S, T\subseteq (\omega^{\omega})^2\times \omega^2$ so that $$R(n,m)\Leftrightarrow \exists f \forall g S(f,g,n,m), \mbox{ and}$$ $$\neg R(n,m)\Leftrightarrow \exists f \forall g T(f,g,n,m).$$ Define $\Pi^1_1$ sets $$R_0=\{(h,\langle n,m\rangle)\mid h(0)=0\wedge \exists f\forall g (S(f,g,n,m)\wedge \forall n(f(n)=h(n+1)))\}$$ and $$R_1=\{(h,\langle n,m\rangle)\mid h(0)=1\wedge \exists f\forall g (T(f,g,n,m)\wedge \forall n(f(n)=h(n+1)))\}.$$ By $\Pi^1_1$-uniformization Theorem, they both can be uniformized by $\Pi^1_1$ partial functions $p_{R_0}:\omega\to \omega^{\omega}$ and $p_{R_1}:\omega\to \omega^{\omega}$. Let $p=p_{R_0} \cup p_{R_1}$. Then $p$ is a $\Pi^1_1$ total function and can viewed as a $\Pi^1_1$-singleton. Then $R$ is recursive in $p$ and so $\alpha<\omega_1^p<\delta$.

Thus $\delta^1_2=\delta$.

share|cite|improve this answer
Thanks a lot! It is a nice proof. – Wei Wang Jul 22 '12 at 11:07

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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