CMPSCI 401: Theory of Computation

Final Exam, Spring 2012

David Mix Barrington

7 May 2012

Directions:

• Answer the problems on the exam pages.
• There are nine problems for 125 total points. Actual scale was A = 100, C = 65.
• If you need extra space use the back of a page.
• No books, notes, calculators, or collaboration.
• The first six questions are true/false, with five points for the correct boolean answer and up to five for a correct justification of your answer -- a proof, counterexample, quotation from the book or from lecture, etc. -- note that there is no reason not to guess if you don't know.

  Q1: 10 points
  Q2: 10 points
  Q3: 10 points
  Q4: 10 points
  Q5: 10 points
  Q6: 10 points
  Q7: 20 points
  Q8: 15 points
  Q9: 30 points
 Total: 125 points

Here are some languages and definitions used on the exam:

• A one-way input Turing machine (OWITM) is a deterministic Turing machine that has a read-only input tape on which the head may never move left. It may have other tapes as specified, and these (if any) are normal read-write tapes with normal heads.
• An alternating finite automaton (AFA) M has the normal components of an NFA, including a transition function δ: (Q × Σ) → P(Q) (where P(Q) is the power set of Q). In addition the states are divided into "White" and "Black" states. The language L(M) is defined in terms of a two-player game as follows. If w is the input string, the two players White and Black decide what choices M will make as it reads w, with White choosing the move from White states and Black from Black states. White wins the game, and thus w ∈ L(M), if and only if she has a winning strategy that will always leave M in a final state after reading w, whatever choices Black makes.
• The language A_DFA is the set of pairs (M, w) such that M is a DFA and w ∈ L(M), and A_NFA is the analogous language for NFA's.
• The language TSP is the set of all pairs (G, t) where G is a directed graph with positive integer labels given in binary, and there exists a path in G, of total weight at most t, that includes each vertex exactly once.
• The language 3-TSP is the subset of TSP where each edge label is either 1, 2, or 3.
• The language EXACT-PATH is the set of all 4-tuples (G, x, y, t) such that G is a directed graph with positive integer labels, given in binary, and there exists a path from x to y in G, with total weight exactly t.
• The following languages may be assumed without proof to be NP-complete, as they were proved to be so either in Sipser or on the homework:
  • A_NP = {(M, w, 1^t): M is a nondeterministic TM and it is possible for M to accept w in t steps}
  • SAT or CNF-SAT is the set of boolean formulas in conjunctive normal form that have at least one input setting making them true.
  • 3-SAT is the seth set of boolean formulas in CNF, with all clauses of size 3, that have at least one input setting making them true.
  • CLIQUE = {(G, k): G is an undirected graph and there exists a set X of k vertices in G such that every pair of distinct vertices in X has an edge}
  • VERTEX-COVER = {(G, k): G is an undirected graph and there exists a set X of k vertices in G such that every edge in G has at least one endpoint in X.}
  • DOMINATING-SET = {(G, k): G is an undirected graph and there exists a set X of k vertices in G such that every vertex in G is either in X or has an edge to a vertex in X}
  • HAM-PATH is the set of directed graphs G such that there exists a path in G that visits each vertex exactly once.
  • UHAMPATH is the set of undirected graphs G such that there exists a path in G that visits each vertex exactly once.
  • 3-COLOR is the set of undirected graphs G such that the vertices of G can each be assigned one of three colors, so that no edge connects two vertices of the same color.
  • SUBSET-SUM is the set of sequences of positive integers (s₁, ..., s_m, t) such that there exists a subset X of {1, ..., m} such that the sum of s_i, for all i in X, equals t.

• Let S = (s₁, ..., s_n) be any sequence of positive integers. We define a labeled directed graph G_S with 3n + 1 nodes and 4n edges as follows. The nodes are a₀ together with a_i, b_i, and c_i for each i with 1 ≤ i ≤ n. For each such i, there is an edge from a_i-1 to b_i with label 1 + s_i. All other edges have label 1 -- for each i there are edges from a_i-1 to c_i, from b_i to a_i, and from c_i to a_i. Here is a picture of G_S where S is the sequence (6, 2, 3) -- edges with no labels are assumed to be labeled with 1:

  
  
      7
  a0 ---> b1
   |      |
   |      |
   V      V   3
  c1 ---> a1 ---> b2
          |       |
          |       |
          V       V   4
          c2 ---> a2 ---> b3
                  |       |
                  |       |
                  V       V
                  c3 ---> a3
  

• Question 1 (10): True or false with justification: The language A_NFA is poly-time reduciable to A_DFA, that is, A_NFA ≤_P A_DFA.

• Question 2 (10): True or false with justification: There does not exist a context-free language X such that A_DFA ≤_L X, that is, such that A_DFA is log-space reducible to X.

• Question 3 (10): True or false with justification: Assuming that P ≠ NP, the language A_NFA is NP-complete.

• Question 4 (10): True or false with justification: Let h be the mapping that takes a sequence of positive integers S, given in binary, to the labeled directed graph G_S as defined above, given as an adjacency list with binary labels. Then h can be computed by a log-space transducer.

• Question 5 (10): True or false with justification: Let M be a one-take OWITM as defined above. (So M's only tape is its one-way read-only input tape.) Then L(M) is a context-free language.

• Question 6 (10): True or false with justification: Let M be a k-tape OWITM for some positive integer k, so that k - 1 of the tapes are ordinary ones. Then there is a two-tape OWITM M' such that L(M) = L(M').

• Question 7 (20): These question deals with AFA's (alternating finite automata) as defined above.
  • (a, 10): Let M be an AFA with k states, let w be any string over the input alphabet of M, and define a function f_w from P(Q) to {0, 1} as follows. (Recall that P(Q) is the power set of the state set Q, that is, P(Q) = {S: S ⊆ Q}.) For any set S, consider the game played on M with input w whre White wants to leave M in some state in S at the end and Black wants the state to not be in S. Then f_w(S) = 1 if and only if White has a winning strategy for this game. Prove that if for two strings u and v, the functions f_u and f_v are the same, then for any string z, the strings uz and vz are either both in L(M) or both not in L(M).

  • (b, 10): Using the result of part (a), whether you proved it or not, prove that L(M) is a regular language. If we use this result to construct a DFA equivalent to M, how many states does the DFA have?

• Question 8 (15): Let NEVER-LEFT be the set of pairs (M, w) such that M is a deterministic Turing machine with a read-only input tape (and any constant number of ordinary tapes), w is a string in M's input alphabet, and M accepts w without ever moving left on its input tape. Prove that the language NEVER-LEFT is Turing recognizable but not Turing decidable. (Note added during test: do not make any assumptions about the machine M other than those you are given.)

• Question 9 (30): Here are two NP-completeness proves for languages defined above. You will probably want to choose from the given NP-complete languages there for your reductions.
  • (a, 15) Prove that the language 3-TSP is NP-complete.

  • (b, 15) Prove that the langauge EXACT-PATH is NP-complete. (Hint: Consider the definition of the graph G_S for a sequence S, given above.)

  Last modified 20 May 2012