2
$\begingroup$

$A$ is a set of all $\langle M \rangle$ that $M$ is a TM halting on all input strings $w$ such that $\lvert w \rvert \le q(M)$ where $q(M)$ is the number of states in $M$.

Is $A$ semi-decidable? Is a complement of $A$ semidecidable?

I think $A$ is semi-decidable. We can construct $M1$.

$M1$ = "On input $\langle M \rangle$ where $M$ is a TM

Simulate $M$ on input with all of the string whose length is less than $q(M)$. If it halts for all, accept"

The complement of $A$ is not semi-decidable. But I'm not sure how to prove it.

$\endgroup$

2 Answers 2

0
$\begingroup$

To show that $\overline{A}\notin RE$ you want to show that another language $B\notin RE$ is reducible to it.

First, the complement of $A$ is the set of all $\left< M \right>$, where $M$ is a Turing machine, and $\exists w$, $|w|<q(M)$ s.t. $M$ doesn't halt on $w$.

Remember that you can add unreachable states to a Turing machine, therefore increasing $q(M)$. The next step would be picking a language who is known to be not decidable (e.g. the complementary of the halting language $\overline{H}$), and think of a reduction to $\overline{A}$.

$\endgroup$
0
$\begingroup$

The logical idea is take $M, w$ for the halting problem (here in the sense halt is accepting), and manufacture $M'$ that only accepts $w$, and has more than $\lvert w \rvert$ states, if $M$ accepts $w$.... but that is easy, just checking that the input is $w$ takes that many states. If we could decide if $M'$ accepts any string of length at most its number of states, we'd be deciding the halting problem.

$\endgroup$

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.