This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: CS 154 Intro. to Automata and Complexity Theory Handout 26 Autumn 2009 David Dill November 10, 2009 Solution Set 5 Problem 1 (a). One way to see that 4TASAT is in NP consists of guessing four different satisfying assign ments for the variables in the input F , and then checking whether they are indeed satisfying and different, which can be done in polynomial time. Another solution consists of building the following nondeterministic TM. First, compute a satisfying assignment for F (if no one exists, answer NO). Then, compute the boolean formula F 2 : F S , where S is the con junction of the values in the satisfying assignment for F . Repeat the same procedure three more times. If we succeed, say YES. Otherwise say NO. Since a satisfying assignment for a boolean formula can be computed in polynomial time by a nondeterministic TM, so can the solution of 4TASAT. (b). Let X 1 ,...,X n be the variables in F and Y 1 ,Y 2 be two new variables. One possible reduction generates the formula G : F ( Y 1 Y 1 ) ( Y 2 Y 2 ). This can be computed in polynomial time. (c). Let k be the number of different satisfying assignments of F . Each of them can be extended to a different satisfying assignment for G by picking all possible truth values for Y 1 and Y 2 . Therefore G has 4 k different satisfying assignments. In particular, if F is satisfiable (i.e. k 1) then G has at least four satisfying assignments (i.e. 4 k 4). For the other direction, if G has at least four satisfying assignments, then in particular it has one, from which we can extract a satisfying assignment for F (simply by ignoring the values for Y 1 and Y 2 ) and conclude that F is satisfiable. Problem 2 First we show that DS is in NP. We nondeterministically guess a set of nodes S of size k , a candidate to be a dominating set. Now, for each node in the graph we check whether it belongs to S , or else whether some edge connects it to some other node in S . This checking operation takes polynomial time. Now, to show that DS is NPhard we use the following reduction from 3SAT: Given a formula F ( X 1 ,... ,X n ) in 3CNF, with clauses C 1 ...C m , we build the following graph G . For each variable X j we introduce three nodes, denoted x j , x j and y j , connected to each other. This is called the triangle gadget . For each clause C i we introduce an additional node c i . If the clause C i contains the literal X j then we add an edge from c i to the node x j . If the literal is of the form X j then we connect c i to x j . The total number of nodes in G is 3 n + m , while the total number of edges is 3 n +3 m . The reduction is completed by generating....
View Full
Document
 '08
 Motwani,R

Click to edit the document details