4

Assuming P!=NP, there is a result that there are decision problems intermediate between P and NP-complete. That is, the class NP cannot be a union of two disjoint subsets: P and NP-complete.

I could never quite understand the proof of the above result. The proof I saw in a textbook was starting with the assumption that one can enumerate all P and NP-hard problems, and then proceeding with a construction of a function that didn't fit in either. However, this construction seemed a bit fishy to me; in particular, the assumption that one can start with enumerated set of problems in a particular class, the NP.

Could you refer me to a clear self-contained proof of the statement in the 1st paragraph? More generally, what would be a good reference for proofs of such results?

Raphael
  • 72,336
  • 29
  • 179
  • 389
Michael
  • 580
  • 4
  • 10

1 Answers1

4

The result you're describing is called 'Ladner's Theorem.' The Wikipedia article on 'NP-Intermediate' is probably a good place to start - you can find references to Ladner's original paper there.

There's also an extensive list of such problems on the cstheory Stackexchange.

For a couple of proofs of Ladner's Theorem, check out this note adapted from Downy & Fortnow's paper 'Uniformly Hard Languages'.

Community
  • 1
rphv
  • 1,624
  • 12
  • 25