Problem Solving in Automata, Languages, and Complexity

(Dana P.) #1
6.4 Nondeterministic Turing Machines 311

Proof NP C - NPSPACE = PSPACE.^0

We now apply Savitch’s theorem to get some weak space hierarchy theorems

for NSPA CE complexity classes (see also Exercise 6 of this section).

Example 6.30 Show that NSPACE(n) $ NSPACE(n210gn).

Proof By Savitch’s theorem, NSPACE(n) - C DSPACE(n2). By the space

hierarchy theorem,


DSPA CE (n2) $ DSPACE(n210gn).

Moreover, it is obvious that DSPACE(n2 log n) C - NSPACE(n2 log n). There-

fore,


NSPACE(n) $ NSPACE(n2 logn). cl

Lemma 6.31 Let q(n), sz(n), and f(n) be fully space-constructible func-

tions with s2(n) > - n and f(n) > - n. Then,

implies

NSPACE(sl(n)) C - NSPACE(s:!(n))

NSPACJqSl (f(n))) c NSPACJqs2(f (4)).

Proof. For any L E NSPACE(sl (f(n))), define


where $ is a symbol not in the alphabet of L. Then, it is obvious that L’ is

in NSPACE(sl(n)). By our assumption, L’ E NSPACE (s2 (n)). This means

that there exists an NTM A42 with space bound 52(n) accepting L’. Now,

construct a two-worktape NTM M3 as follows:

(1) On input X, MS simulates an f (n)-space marking DTM to create the


string ~$-f(l~l~-l~l on the second tape.

(2) MS then simulates M2 using the second tape as the input tape and the

third tape as the work tape of M2. MS accepts the input x if and only

if M2 accepts ~$f@l)-l~l.

It is clear that MS accepts L in space bound s2( f (n)). (Note: Since sz(n) > - n,
the space used in the second tape is bounded by s2 (f(n)) .) cl


Example 6.32 Show that NSPACE(n) s NSPACE(n1*5).

Proof. By way of contradiction, suppose that NSPACE(n5) C - NSPACE(n).

Choose f(n) = n6 and n4, and apply Lemma 6.31 to the above relation. Then,
we get


NSPACE(n’) C - NSPACE(n6) C - NSPACE(n”).
Free download pdf