8
$\begingroup$

From wikipedia: https://en.wikipedia.org/wiki/Absoluteness_(logic)#Shoenfield's_absoluteness_theorem

"Shoenfield's theorem shows that if there is a model ZF in which a given $\Pi^1_3$ statement $\phi$ is false, then $\phi$ is also false in the constructible universe of that model."

The problem is that the only proofs I can find seem to use "a model of ZF+DC" or "a model of ZFC".

Is wikipedia right? If so, is there a proof from just "a model of ZF" online or short enough to sketch here?

Improve this question
$\endgroup$
1
  • $\begingroup$ Either you meant "true" instead of (both occurrences of) "false" or you meant $\Sigma^1_3$ instead of $\Pi^1_3$. "All reals are constructible" is $\Pi^1_3$ and is false in some models but true in the constructible submodel. $\endgroup$
    – Andreas Blass
    Commented May 24 at 15:46

1 Answer 1

Reset to default
15
$\begingroup$

Wikipedia is correct in that the Shoenfield Absoluteness Theorem holds for plain ZF.

Since the proof of the theorem relies heavily on the absoluteness of well-foundedness, it is tempting to assume DC. However, since the trees that occur in the usual proof of the theorem are canonically well-ordered, DC is not necessary to prove that the well-foundedness of these trees is absolute. For a different approach, see the proof given by Barwise and Fisher in The Shoenfield Absoluteness Lemma. [Israel J. Math. 8 1970, 329-339, MR278934]

Improve this answer
$\endgroup$
2
  • $\begingroup$ Does it somehow follow easily from Theorem 4a in that paper? I don't quite see it. $\endgroup$
    – user5810
    Commented May 9, 2010 at 6:58
  • $\begingroup$ It actually follows from Theorem 3a, see Remark 3 which follows that theorem. $\endgroup$
    – François G. Dorais
    Commented May 18, 2010 at 22:25

You must log in to answer this question.