# Intutive description of the PCP Theorem

The PCP theorem states that:

Every choice trouble in NP has probabilistically checkable proofs of constant query complexity and also logarithmic randomness intricacy.

Can any person offer an instinctive description of just how this can be done?

**Hyperlinks**

- This is an adhere to up from this question.
- There is an evidence here.

A thorough description can be located in several areas. I'll to give an instinctive one.

By Cook-Levin theorem, the Boolean satisfiability problem is NP-complete - ie. every choice trouble in NP can be lowered to it. We will certainly ask of our prover to provide the input and also result of every gateway in the circuit and also consider this as the evidence that $x\in L$.

If we were to stop there, after that the prover can rip off by transforming a solitary little bit in his evidence, which would certainly need us to examine a huge (non - constant) variety of little bits in it. So we have to ask something extra from the prover. This something extra will certainly be the encoding of his evidence making use of some (unique) mistake - dealing with code. With ease, this "spots" the incorrect little bits onto a lot of little bits in the code - word.

The verifier obtains a word from the prover, there are 2 methods which the prover could attempt and also rip off, this word could not be a code - word in the picked code, or it could be the encoding of something which isn't an evidence. We'll check out the (a little various) splitting up of instances of words either being much from every codeword (a lot of little bits have to be transformed to get to a codeword) or that it is close to codeword which isn't an encoding of an evidence.

To be able to identify these we require that our code has the adhering to 2 buildings : 1) in your area checkable - we can, by reviewing a constant variety of little bits, identify w.h.p if a word is much from any kind of codeword. 2) in your area decodable - we can, by reviewing a constant variety of little bits, translate w.h.p a little bit from the inscribed word. (searching for such codes is hard, hadamard has this buildings yet the code - word's dimension is rapid, RM codes have these buildings too (to a minimal level) and also they are the ones usually made use of.

So the verifier checks if words is close to being a code - word (making use of 1), and also if the examination does well, selects an arbitrary gateway and also translates it's results and also result (making use of 2). this suffices to attain constant question intricacy and also logarithmic randomness.

It needs to be stated that (Dinur) has a combinatorial evidence of PCP which is fairly various from what i've discribed, yet her variation is (for me, at the very least) much less instinctive.

Related questions