What is complexity theoretic implication of following possibilities - $NP=coNP=BPP=RP$ or $coNP\neq NP=BPP=RP$ (consensus is these seem impossible)?
Asked By : Turbo
Answered By : Luke Mathieson
Unless I'm missing something obvious (it is getting late here), the second isn't possible - $BPP = coBPP$, so $NP=BPP \rightarrow NP = coBPP \rightarrow NP = coNP$.
In the first case, some of the consequences would be:
- "Feasible" algorithms for $NP$-complete problems, for at least the randomised version of feasible.
- $PH \subseteq NP$
- Unless $P = NP$, adding randomness adds computational power.
- Strong pseudorandom number generators do not exist (nor do one way functions). (we would need $BPP = P$ for them to exist, but that would imply $P=NP$, in which case they don't).
Best Answer from StackOverflow
Question Source : http://cs.stackexchange.com/questions/40976
0 comments:
Post a Comment
Let us know your responses and feedback