6
$\begingroup$

Question: Suppose $\{a_n : n < \omega \}$ is a $<_T$-ascending sequence in $2^{\omega}$. Can we find $x, y \in 2^{\omega}$ such that for every $z \in 2^{\omega}$, the set of reals computable from each one of $x \oplus z$ and $y \oplus z$ is the Turing ideal generated by $\{a_n : n < \omega\} \cup \{z\}$? So taking $z$ to be computable, this implies that $x, y$ form an exact pair for the Turing ideal generated by $\{a_n : n < \omega\}$.

My thoughts: The usual construction of exact pair has only countably many splitting requirements ($\Phi^{x} = \Psi^{y} = w \implies (\exists n)(w \leq_T a_n)$) but this construction would require continuum many. I suspect such pairs don't exist but I don't have a proof.

$\endgroup$
1
  • 1
    $\begingroup$ I suspect that for $g$ sufficiently Cohen generic, the real $z=(x\Delta g)\oplus (y\Delta g)$ will fail to compute $x$ (or $y$) even from finitely many of the $a_i$s; since $z\oplus x$ and $z\oplus y$ each compute $x\oplus y$, this would give the expected negative answer. But I don't yet see how to show this (in particular, the naive forcing analysis just gets us $(x\Delta y)\oplus a_1\oplus ...\oplus a_n\equiv_Tx\oplus y$ for some $n$, which is entirely possible). $\endgroup$ May 20, 2019 at 9:15

1 Answer 1

5
$\begingroup$

You could look at Theorem 3.1 of this paper:

On the Σ2-Theory of the Upper Semilattice of Turing Degrees Author(s): Carl G. Jockusch, Jr. and Theodore A. Slaman Source: The Journal of Symbolic Logic, Vol. 58, No. 1 (Mar., 1993), pp. 193-204 Published by: Association for Symbolic Logic

3.1 characterizes the possible upper-semi-lattice end-extensions of a countable ideal that can be generated by the ideal and a single new element. Basically, anything that is consistent as an upper-semi-lattice can be realized. In particular, for any $x$ and $y$ as above, there is a $z$ such that $z\oplus x$ and $z\oplus y$ are Turing equivalent, but for all $a_n$, $z\oplus a_n$ computes neither $x$ nor $y$.

Your instincts were correct: there is no pair of the described type.

$\endgroup$

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.