8

I was pondering the fact that maybe the classical hard complexity-theoretic questions are undecidable, not because they are so themselves, but because some set-theoretic foundations makes the complexity-theoretic foundations shaky.

My thoughts was that perhaps something like the Continuum hypothesis makes P vs NP undecidable. So my question is, is there a "finitary" or otherwise obviously sane environment for complexity theory that would discount this theory immediately? I'm aware of simpler structures where P vs NP has been decided, but I don't know how that would fit in.

I apologize in advance if this doesn't make sense.

kastberg
  • 123
  • 1
    A related question that have led to many answers is http://mathoverflow.net/questions/50023/independence-of-p-np – Gil Kalai Apr 07 '11 at 18:46
  • Nice catch, also the paper linked to there, http://www.scottaaronson.com/papers/pnp.pdf seems very useful. – kastberg Apr 07 '11 at 19:31

1 Answers1

7

The statement that P=NP can be expressed in first-order arithmetic, and that part of mathematics is unaffected by the known methods of proving set-theoretic independence results (forcing, inner models).

Andreas Blass
  • 72,290
  • 4
    Andreas: Though it is true that something like CH bears no effect on these problems, there are also techniques for producing non-standard models, such as the methods developed by Harvey Friedman. They were devised so we can obtain, for example, independence of certain arithmetic statements from ZFC. – Andrés E. Caicedo Apr 07 '11 at 19:04
  • I think it is possible than P=NP is independent of ZFC. There are no obvious obstructions to this statement. –  Apr 08 '11 at 01:42