help with problems on computability needed!!
- From: ahtwong@xxxxxxxxx
- Date: 15 Jun 2005 20:12:34 -0700
Hi all, I was wondering if anyone would be able to help me with the
following problems. I'm really bad at these... that's why I have to
retake this course on computational theory and complexity... I could
use any help I can get!
First question:
Consider a table whose rows are labelled with TMs, whose columns are
labelled with TM encodings, and where the entry at (Mi, <Mj>) contains
a 1 if Mi accepts <Mj> in at most |<Mj>|+t steps, and a 0 otherwise.
Use a diagonalization argument to show that for every t, there exists a
language L which is decidable in general, but not decidable with lag t.
We say that a language L is decidable with lag t if there exists a TM M
deciding L, which also satifies the requirement that on every input w,
M halts in at most |w| + t steps.
This second question is more tricky, I think:
Let M1,M2, . . . ,Mi, . . . be the list of all TMs in lexicographical
order.
For every positive integer k, define f(k) to be the index in the above
list of the k-th TM M such that L(M) = NULL. Formally,
f(k) = max{i : there exists J, which is a subset of {1 . . . i - 1}
such that (|J| = k - 1 and for all j belonging to {1 . . . i - 1},
L(Mj) = NULL iff j belongs to J)}
Notice that f is well defined for every k, since there are infinitely
many TMs M with L(M) = NULL. For example, if L(M2) = L(M5) = NULL and
L(M1),L(M3),L(M4) are not empty, then f(1) = 2 and f(2) = 5.
Prove that f is not computable.
I hope these aren't too difficult, although they are making me scratch
my head! Thanks for all your help, guys!!!
.
- Follow-Ups:
- A little advice on strategy (no solutions. Sorry)
- From: Theory Man
- A little advice on strategy (no solutions. Sorry)
- Prev by Date: shortest path with constraints that some nodes can not be on the same pth
- Next by Date: Re: ZFC
- Previous by thread: shortest path with constraints that some nodes can not be on the same pth
- Next by thread: A little advice on strategy (no solutions. Sorry)
- Index(es):
Relevant Pages
|