0
$\begingroup$

If $L\subseteq \{0, 1\}^∗$ is a language, then we denote by $\overline{L}$ the complement of $L$. That is, $\overline{L} = \{0, 1\}^∗\setminus L.$ We make the following one definition of $\mathrm{coNP}$: $$\mathrm{coNP} = \{L : \overline{L} ∈ \mathrm{NP}\}.$$

$\mathrm{coNP}$, ALTERNATIVE DEFINITION:

For every $L\subseteq \{0, 1\}^∗$, we say that $L\in \mathrm{coNP}$ if there exists a polynomial $p \colon \mathbb{N}\to \mathbb{N}$ and a polynomial-time TM $M$ such that for every $x\in \{0, 1\}^∗$, $$x\in L \iff \forall u \in \{0, 1\}^{p(|x|)}\: M(x, u) = 1.$$

I am not able to understand why and how both two definitions are equivalent; the book mentions these two but does not say anything about the equivalence.

$\endgroup$
1
  • $\begingroup$ Oh, I have checked the draft version, it's a erratum. You thus should find the published version on libgen, it have been revised partially, including the definition of coNP here. It's so sad that I have not yet found the official errata page. $\endgroup$ Commented Nov 22 at 5:16

1 Answer 1

1
$\begingroup$

We should prove that the former implies the latter definition and vice versa.

Let $\mathbf{coNP}=\{L:\overline{L}\in\mathbf{NP}\}$. For $L\in\mathbf{coNP}$, there exists a polynomial $p(n)$ and a polynomial-time TM $M$ such that for every $$x\in\overline{L}\iff\exists u\in\{0,1\}^{p(n)}\;M(x,u)=1$$ because $\overline{L}\in\mathbf{NP}$. Thus, $$x\in L\iff\forall u\in\{0,1\}^{p(n)}\;M(x,u)=0.$$ We additionally construct a polynomial-time deterministic TM $M'$ s.t. $M'(x)=1-M(x), \forall x\in\{0,1\}^*$. We finally have $L\in\mathbf{coNP}$, there exists a polynomial $p(n)$ and a polynomial-time TM $M'$ such that for every $$x\in L\iff\forall u\in\{0,1\}^{p(n)}\;M'(x,u)=1.$$

The other, i.e. the latter implies the former, is similar.

$\endgroup$
2
  • $\begingroup$ Would you edit your answer based on how the latter implies the former? $\endgroup$
    – Beduin
    Commented Dec 1 at 23:33
  • $\begingroup$ @Beduin It is similar, you should try to do it! $\endgroup$ Commented Dec 3 at 4:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.