Section 7.4

More NP-Complete Problems

[12/58%]
Show
LaTeX
Example 7.36

Does the problem VC remain NP-complete if each edge is to be covered by exactly one vertex?

?
Example 7.37

Prove that MCG is NP-complete.

?
Example 7.38

Partition is NP-complete.

?
Example 7.39

Knapsack is NP-complete.

?
Example 7.40

BP is NP-complete.

?
Example 7.41

Show that the following problem is NP-complete: Given positive integers a1,…,ana_{1}, \ldots , a_{n}, determine whether

∫02π(∏i=1ncos⁡(ait))dt≠0 \int _{0}^{2 \pi }\left(\prod _{i=1}^{n} \cos \left(a_{i} t\right)\right) d t \neq 0
?
Example 7.42

SMT is NP-complete.

?
Problem 7.4.1

For each of the following problems, determine whether it is NP\mathrm{NP}-complete or is in PP. If it is NP\mathrm{NP}-complete, find a reduction from a known NP\mathrm{NP} complete problem to it. If it is in PP, find a polynomial-time algorithm for it.

?
(a)

Given a graph GG, two vertices ss and tt in GG and a positive integer kk, determine whether there is a path between ss and tt of length less than or equal to kk.

(b)

Given a graph GG, two vertices ss and tt in GG and a positive integer kk, determine whether there is a path between ss and tt of length greater than or equal to kk.

(c)

Given a directed graph GG, determine whether GG has a cycle of an odd length.

(d)

Given a directed graph GG, determine whether GG has a cycle of an even length.

(e)

Given a graph GG, determine whether GG has a cycle of an odd length.

(f)

Given an edge-weighted complete graph GG, three subsets X1,X2X_{1}, X_{2}, X3X_{3} of vertices and a positive integer kk, determine whether there is a subgraph HH of GG of weight ≤k\leq k that contains a spanning tree for each of X1,X2X_{1}, X_{2} and X3X_{3}. (A spanning tree for a subset X⊆VX \subseteq V is a connected subgraph HH of GG on the vertex set XX that has no cycle.)

(g)

Given a graph G=(V,E)G=(V, E) and a positive integer kk, determine whether there is a subset T⊆ET \subseteq E of at most kk edges such that every vertex in VV is incident with at least one edge in TT.

(h)

Regular Expression Nonequivalence: Given two regular expressions r1r_{1} and r2r_{2} without the Kleene-star operation, determine whether L(r1)≠L(r2)L\left(r_{1}\right) \neq L\left(r_{2}\right).

(i)

Bounded PCP (see Exercise 7(d) of Section 7.1).

(j)

Bounded Tiling (see Exercise 7(e) of Section 7.1).

Problem 7.4.2

Prove that the following variations of the problem VC are NP\mathrm{NP}-complete:

?
(a)

Planar VC: The problem VC restricted to planar graphs. (A planar graph is a graph that can be drawn on the two-dimensional plane such that no two edges cross each other at a non-vertex point.)

(b)

Cubic VC: The problem VC restricted to cubic graphs. (A cubic graph is a graph such that every vertex has degree three.)

(c)

Planar Connected-VC-4: Given a planar graph G=(V,E)G=(V, E) with each vertex in VV having degree at most 4 and an integer k>0k>0, determine whether there is a vertex cover CC of GG of size kk such that the induced subgraph G∣C\left.G\right|_{C} on vertex set CC is connected.

(d)

Given a graph GG, determine whether GG has a vertex cover CC satisfying the following conditions:

  1. The subgraph G∣C\left.G\right|_{C} induced by CC has no isolated point.

  2. Every vertex in CC is adjacent to a vertex not in CC.

Problem 7.4.3

The polar representation of a CNF FF is a bipartite graph GF=(V1,V2,E)G_{F}= \left(V_{1}, V_{2}, E\right), where V1V_{1} is the set of all variables and V2V_{2} is the set of all clauses in FF, and there is an edge in EE between xi∈V1x_{i} \in V_{1} and cj∈V2c_{j} \in V_{2} if and only if the variable xix_{i} appears in the clause cjc_{j} (in the form of xix_{i} or xˉi\bar{x}_{i}). Prove the following statements:

?
(a)

(x+y+zˉ)(xˉ+z+w)(xˉ+z+wˉ)(yˉ+z+u)(yˉ+z+uˉ)(x+y+\bar{z})(\bar{x}+z+w)(\bar{x}+z+\bar{w})(\bar{y}+z+u)(\bar{y}+z+\bar{u}) has a planar polar representation and it is satisfiable if and only if x+y=zx+y=z.

(b)

There exists a 3-CNF formula FF with a planar polar representation and three variables x,yx, y, and zz, such that FF is satisfiable if and only if xy=zx y=z.

(c)

There exists a 3-CNF formula FF with a planar polar representation and three variables x,yx, y, and zz, such that FF is satisfiable if and only if x⊕y=zx \oplus y=z. [Hint: x⊕y=x(xˉ+yˉ)+(xˉ+yˉ)yx \oplus y=x(\bar{x}+\bar{y})+(\bar{x}+\bar{y}) y.]

(d)

The following problem, called Planar Polar-3Sat, is NPcomplete: Given a 3-CNF formula FF with a planar polar representation, determine whether FF is satisfiable or not. [Hint: Applying the fact that x⊕(x⊕y)=yx \oplus (x \oplus y)=y and (x⊕y)⊕y=x(x \oplus y) \oplus y=x to the construction.]

Problem 7.4.4

The nonpolar representation of a CNF FF is a graph GF=(V,E)G_{F}=(V, E), with the vertex set VV consisting of all literals and all clauses in FF, and the edge set EE consisting of all pairs {x,xˉ}\left\{ x, \bar{x}\right\} over all variables xx in FF plus all literal-clause pairs {z,c}\left\{ z, c\right\} such that zz occurs in cc. Prove the following statements:

?
(a)

If the nonpolar representation of a CNF formula FF is planar, then its polar representation must also be planar. However, the converse is not necessarily true.

(b)

The following problem, called Planar Nonpolar-3Sat, is NPcomplete: Given a 3-CNF formula FF with a planar nonpolar representation, determine whether FF is satisfiable or not.

Problem 7.4.5

A (k,ℓ)−CNFF(k, \ell )-C N F F is a CNF such that each clause contains exactly kk literals and each variable occurs in at most ℓ\ell clauses. The problem (k,ℓ)(k, \ell )-Sat is the problem 3SaT restricted to (k,ℓ)(k, \ell )-CNF's. Prove the following results:

?
(a)

For any k>0k>0, every (k,k)(k, k)-CNF FF is satisfiable.

(b)

For any ℓ>0\ell >0, (2, ℓ\ell)-SAT is polynomial-time solvable. (In fact, the problem 2Sat, which is the problem 3Sat restricted to CNF's with at most 2 literals in each clause, is polynomial-time solvable.)

(c)

(3,4)(3,4)-Sat is NP\mathrm{NP}-complete.